fix: sort Nous Portal model list (opus first, sonnet lower)

fetch_nous_models() returned models in whatever order the API gave them, which put sonnet near the top. Add a priority sort so users see the best models first: opus > pro > other > sonnet.
fix: Docker backend fails when docker is not in PATH (macOS gateway)
2026-03-10 23:20:46 -07:00 · 2026-03-10 20:45:13 -07:00 · 2026-03-10 17:12:34 -07:00 · 2026-03-10 15:59:08 -07:00 · 2026-03-10 15:29:24 -07:00 · 2026-03-10 15:22:44 -07:00
395 changed files with 25630 additions and 3149 deletions
--- a/.env.example
+++ b/.env.example
@@ -53,10 +53,6 @@ MINIMAX_CN_API_KEY=
 # Get at: https://firecrawl.dev/
 FIRECRAWL_API_KEY=

-# Nous Research API Key - Vision analysis and multi-model reasoning
-# Get at: https://inference-api.nousresearch.com/
-NOUS_API_KEY=
-
 # FAL.ai API Key - Image generation
 # Get at: https://fal.ai/
 FAL_KEY=
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -1,80 +1,61 @@
 # Hermes Agent - Development Guide

-Instructions for AI coding assistants (GitHub Copilot, Cursor, etc.) and human developers.
-
-Hermes Agent is an AI agent harness with tool-calling capabilities, interactive CLI, messaging integrations, and scheduled tasks.
+Instructions for AI coding assistants and developers working on the hermes-agent codebase.

 ## Development Environment

-**IMPORTANT**: Always use the virtual environment if it exists:
 ```bash
-source venv/bin/activate  # Before running any Python commands
+source .venv/bin/activate  # ALWAYS activate before running Python
 ```

 ## Project Structure

 ```
 hermes-agent/
-├── agent/                # Agent internals (extracted from run_agent.py)
-│   ├── auxiliary_client.py   # Shared auxiliary OpenAI client (vision, compression, web extract)
-│   ├── model_metadata.py     # Model context lengths, token estimation
+├── run_agent.py          # AIAgent class — core conversation loop
+├── model_tools.py        # Tool orchestration, _discover_tools(), handle_function_call()
+├── toolsets.py           # Toolset definitions, _HERMES_CORE_TOOLS list
+├── cli.py                # HermesCLI class — interactive CLI orchestrator
+├── hermes_state.py       # SessionDB — SQLite session store (FTS5 search)
+├── agent/                # Agent internals
+│   ├── prompt_builder.py     # System prompt assembly
 │   ├── context_compressor.py # Auto context compression
 │   ├── prompt_caching.py     # Anthropic prompt caching
-│   ├── prompt_builder.py     # System prompt assembly (identity, skills index, context files)
+│   ├── auxiliary_client.py   # Auxiliary LLM client (vision, summarization)
+│   ├── model_metadata.py     # Model context lengths, token estimation
 │   ├── display.py            # KawaiiSpinner, tool preview formatting
+│   ├── skill_commands.py     # Skill slash commands (shared CLI/gateway)
 │   └── trajectory.py         # Trajectory saving helpers
-├── hermes_cli/           # CLI implementation
-│   ├── main.py           # Entry point, command dispatcher
-│   ├── banner.py         # Welcome banner, ASCII art, skills summary
-│   ├── commands.py       # Slash command definitions + autocomplete
-│   ├── callbacks.py      # Interactive prompt callbacks (clarify, sudo, approval)
+├── hermes_cli/           # CLI subcommands and setup
+│   ├── main.py           # Entry point — all `hermes` subcommands
+│   ├── config.py         # DEFAULT_CONFIG, OPTIONAL_ENV_VARS, migration
+│   ├── commands.py       # Slash command definitions + SlashCommandCompleter
+│   ├── callbacks.py      # Terminal callbacks (clarify, sudo, approval)
 │   ├── setup.py          # Interactive setup wizard
-│   ├── config.py         # Config management & migration
-│   ├── status.py         # Status display
-│   ├── doctor.py         # Diagnostics
-│   ├── gateway.py        # Gateway management
-│   ├── uninstall.py      # Uninstaller
-│   ├── cron.py           # Cron job management
-│   └── skills_hub.py     # Skills Hub CLI + /skills slash command
-├── tools/                # Tool implementations
-│   ├── registry.py            # Central tool registry (schemas, handlers, dispatch)
-│   ├── approval.py            # Dangerous command detection + per-session approval
-│   ├── environments/          # Terminal execution backends
-│   │   ├── base.py            # BaseEnvironment ABC
-│   │   ├── local.py           # Local execution with interrupt support
-│   │   ├── docker.py          # Docker container execution
-│   │   ├── ssh.py             # SSH remote execution
-│   │   ├── singularity.py     # Singularity/Apptainer + SIF management
-│   │   ├── modal.py           # Modal cloud execution
-│   │   └── daytona.py         # Daytona cloud sandboxes
-│   ├── terminal_tool.py       # Terminal orchestration (sudo, lifecycle, factory)
-│   ├── todo_tool.py           # Planning & task management
-│   ├── process_registry.py    # Background process management
-│   └── ...                    # Other tool files
-├── gateway/              # Messaging platform adapters
-│   ├── platforms/        # Platform-specific adapters (telegram, discord, slack, whatsapp)
-│   └── ...
-├── cron/                 # Scheduler implementation
-├── environments/         # RL training environments (Atropos integration)
-├── skills/               # Bundled skill sources
-├── optional-skills/      # Official optional skills (not activated by default)
-├── cli.py                # Interactive CLI orchestrator (HermesCLI class)
-├── hermes_state.py       # SessionDB — SQLite session store (schema, titles, FTS5 search)
-├── run_agent.py          # AIAgent class (core conversation loop)
-├── model_tools.py        # Tool orchestration (thin layer over tools/registry.py)
-├── toolsets.py           # Tool groupings
-├── toolset_distributions.py  # Probability-based tool selection
+│   └── skin_engine.py    # Skin/theme engine — CLI visual customization
+├── tools/                # Tool implementations (one file per tool)
+│   ├── registry.py       # Central tool registry (schemas, handlers, dispatch)
+│   ├── approval.py       # Dangerous command detection
+│   ├── terminal_tool.py  # Terminal orchestration
+│   ├── process_registry.py # Background process management
+│   ├── file_tools.py     # File read/write/search/patch
+│   ├── web_tools.py      # Firecrawl search/extract
+│   ├── browser_tool.py   # Browserbase browser automation
+│   ├── code_execution_tool.py # execute_code sandbox
+│   ├── delegate_tool.py  # Subagent delegation
+│   ├── mcp_tool.py       # MCP client (~1050 lines)
+│   └── environments/     # Terminal backends (local, docker, ssh, modal, daytona, singularity)
+├── gateway/              # Messaging platform gateway
+│   ├── run.py            # Main loop, slash commands, message dispatch
+│   ├── session.py        # SessionStore — conversation persistence
+│   └── platforms/        # Adapters: telegram, discord, slack, whatsapp, homeassistant, signal
+├── cron/                 # Scheduler (jobs.py, scheduler.py)
+├── environments/         # RL training environments (Atropos)
+├── tests/                # Pytest suite (~2500+ tests)
 └── batch_runner.py       # Parallel batch processing
 ```

-**User Configuration** (stored in `~/.hermes/`):
- `~/.hermes/config.yaml` - Settings (model, terminal, toolsets, etc.)
- `~/.hermes/.env` - API keys and secrets
- `~/.hermes/pairing/` - DM pairing data
- `~/.hermes/hooks/` - Custom event hooks
- `~/.hermes/image_cache/` - Cached user images
- `~/.hermes/audio_cache/` - Cached user voice messages
- `~/.hermes/sticker_cache.json` - Telegram sticker descriptions
+**User config:** `~/.hermes/config.yaml` (settings), `~/.hermes/.env` (API keys)

 ## File Dependency Chain

@@ -88,698 +69,275 @@ model_tools.py  (imports tools/registry + triggers tool discovery)
 run_agent.py, cli.py, batch_runner.py, environments/
 ```

-Each tool file co-locates its schema, handler, and registration. `model_tools.py` is a thin orchestration layer.
-
 ---

-## AIAgent Class
-
-The main agent is implemented in `run_agent.py`:
+## AIAgent Class (run_agent.py)

 ```python
 class AIAgent:
-    def __init__(
-        self,
-        model: str = "anthropic/claude-sonnet-4.6",
-        api_key: str = None,
-        base_url: str = "https://openrouter.ai/api/v1",
-        max_iterations: int = 60,        # Max tool-calling loops
+    def __init__(self,
+        model: str = "anthropic/claude-opus-4.6",
+        max_iterations: int = 90,
        enabled_toolsets: list = None,
        disabled_toolsets: list = None,
-        verbose_logging: bool = False,
-        quiet_mode: bool = False,         # Suppress progress output
-        tool_progress_callback: callable = None,  # Called on each tool use
-    ):
-        # Initialize OpenAI client, load tools based on toolsets
-        ...
-    
-    def chat(self, user_message: str, task_id: str = None) -> str:
-        # Main entry point - runs the agent loop
-        ...
+        quiet_mode: bool = False,
+        save_trajectories: bool = False,
+        platform: str = None,           # "cli", "telegram", etc.
+        session_id: str = None,
+        skip_context_files: bool = False,
+        skip_memory: bool = False,
+        # ... plus provider, api_mode, callbacks, routing params
+    ): ...
+
+    def chat(self, message: str) -> str:
+        """Simple interface — returns final response string."""
+
+    def run_conversation(self, user_message: str, system_message: str = None,
+                         conversation_history: list = None, task_id: str = None) -> dict:
+        """Full interface — returns dict with final_response + messages."""
 ```

 ### Agent Loop

-The core loop in `_run_agent_loop()`:
-
-```
-1. Add user message to conversation
-2. Call LLM with tools
-3. If LLM returns tool calls:
-   - Execute each tool
-   - Add tool results to conversation
-   - Go to step 2
-4. If LLM returns text response:
-   - Return response to user
-```
+The core loop is inside `run_conversation()` — entirely synchronous:

 ```python
-while turns < max_turns:
-    response = client.chat.completions.create(
-        model=model,
-        messages=messages,
-        tools=tool_schemas,
-    )
-    
+while api_call_count < self.max_iterations and self.iteration_budget.remaining > 0:
+    response = client.chat.completions.create(model=model, messages=messages, tools=tool_schemas)
    if response.tool_calls:
        for tool_call in response.tool_calls:
-            result = await execute_tool(tool_call)
+            result = handle_function_call(tool_call.name, tool_call.args, task_id)
            messages.append(tool_result_message(result))
-        turns += 1
+        api_call_count += 1
    else:
        return response.content
 ```

-### Conversation Management
-
-Messages are stored as a list of dicts following OpenAI format:
-
-```python
-messages = [
-    {"role": "system", "content": "You are a helpful assistant..."},
-    {"role": "user", "content": "Search for Python tutorials"},
-    {"role": "assistant", "content": None, "tool_calls": [...]},
-    {"role": "tool", "tool_call_id": "...", "content": "..."},
-    {"role": "assistant", "content": "Here's what I found..."},
-]
-```
-
-### Reasoning Model Support
-
-For models that support chain-of-thought reasoning:
- Extract `reasoning_content` from API responses
- Store in `assistant_msg["reasoning"]` for trajectory export
- Pass back via `reasoning_content` field on subsequent turns
+Messages follow OpenAI format: `{"role": "system/user/assistant/tool", ...}`. Reasoning content is stored in `assistant_msg["reasoning"]`.

 ---

 ## CLI Architecture (cli.py)

-The interactive CLI uses:
- **Rich** - For the welcome banner and styled panels
- **prompt_toolkit** - For fixed input area with history, `patch_stdout`, slash command autocomplete, and floating completion menus
- **KawaiiSpinner** (in run_agent.py) - Animated kawaii faces during API calls; clean `┊` activity feed for tool execution results
-
-Key components:
- `HermesCLI` class - Main CLI controller with commands and conversation loop
- `SlashCommandCompleter` - Autocomplete dropdown for `/commands` (type `/` to see all)
- `agent/skill_commands.py` - Scans skills and builds invocation messages (shared with gateway)
- `load_cli_config()` - Loads config, sets environment variables for terminal
- `build_welcome_banner()` - Displays ASCII art logo, tools, and skills summary
- `_preload_resumed_session()` - Loads session history early (before banner) for immediate display on resume
- `_display_resumed_history()` - Renders a compact conversation recap in a Rich Panel on session resume
-
-CLI UX notes:
- Thinking spinner (during LLM API call) shows animated kawaii face + verb (`(⌐■_■) deliberating...`)
- When LLM returns tool calls, the spinner clears silently (no "got it!" noise)
- Tool execution results appear as a clean activity feed: `┊ {emoji} {verb} {detail} {duration}`
- "got it!" only appears when the LLM returns a final text response (`⚕ ready`)
- The prompt shows `⚕ ❯` when the agent is working, `❯` when idle
- Pasting 5+ lines auto-saves to `~/.hermes/pastes/` and collapses to a reference
- Multi-line input via Alt+Enter or Ctrl+J
- When resuming a session (`--continue`/`--resume`), a "Previous Conversation" panel shows previous messages before the input prompt (configurable via `display.resume_display`)
- `/commands` - Process user commands like `/help`, `/clear`, `/personality`, etc.
- `/skill-name` - Invoke installed skills directly (e.g., `/axolotl`, `/gif-search`)
-
-CLI uses `quiet_mode=True` when creating AIAgent to suppress verbose logging.
-
-### Skill Slash Commands
-
-Every installed skill in `~/.hermes/skills/` is automatically registered as a slash command.
-The skill name (from frontmatter or folder name) becomes the command: `axolotl` → `/axolotl`.
-
-Implementation (`agent/skill_commands.py`, shared between CLI and gateway):
-1. `scan_skill_commands()` scans all SKILL.md files at startup, filtering out skills incompatible with the current OS platform (via the `platforms` frontmatter field)
-2. `build_skill_invocation_message()` loads the SKILL.md content and builds a user-turn message
-3. The message includes the full skill content, a list of supporting files (not loaded), and the user's instruction
-4. Supporting files can be loaded on demand via the `skill_view` tool
-5. Injected as a **user message** (not system prompt) to preserve prompt caching
+- **Rich** for banner/panels, **prompt_toolkit** for input with autocomplete
+- **KawaiiSpinner** (`agent/display.py`) — animated faces during API calls, `┊` activity feed for tool results
+- `load_cli_config()` in cli.py merges hardcoded defaults + user config YAML
+- **Skin engine** (`hermes_cli/skin_engine.py`) — data-driven CLI theming; initialized from `display.skin` config key at startup; skins customize banner colors, spinner faces/verbs/wings, tool prefix, response box, branding text
+- `process_command()` is a method on `HermesCLI` (not in commands.py)
+- Skill slash commands: `agent/skill_commands.py` scans `~/.hermes/skills/`, injects as **user message** (not system prompt) to preserve prompt caching

 ### Adding CLI Commands

-1. Add to `COMMANDS` dict with description
-2. Add handler in `process_command()` method
-3. For persistent settings, use `save_config_value()` to update config
-
---
-
-## Hermes CLI Commands
-
-The unified `hermes` command provides all functionality:
-
-| Command | Description |
-|---------|-------------|
-| `hermes` | Interactive chat (default) |
-| `hermes chat -q "..."` | Single query mode |
-| `hermes -c` / `hermes --continue` | Resume the most recent session |
-| `hermes -c "my project"` | Resume a session by name (latest in lineage) |
-| `hermes --resume <session_id>` | Resume a specific session by ID or title |
-| `hermes -w` / `hermes --worktree` | Start in isolated git worktree (for parallel agents) |
-| `hermes setup` | Configure API keys and settings |
-| `hermes config` | View current configuration |
-| `hermes config edit` | Open config in editor |
-| `hermes config set KEY VAL` | Set a specific value |
-| `hermes config check` | Check for missing config |
-| `hermes config migrate` | Prompt for missing config interactively |
-| `hermes status` | Show configuration status |
-| `hermes doctor` | Diagnose issues |
-| `hermes update` | Update to latest (checks for new config) |
-| `hermes uninstall` | Uninstall (can keep configs for reinstall) |
-| `hermes gateway` | Start gateway (messaging + cron scheduler) |
-| `hermes gateway setup` | Configure messaging platforms interactively |
-| `hermes gateway install` | Install gateway as system service |
-| `hermes sessions list` | List past sessions (title, preview, last active) |
-| `hermes sessions rename <id> <title>` | Rename/title a session |
-| `hermes cron list` | View scheduled jobs |
-| `hermes cron status` | Check if cron scheduler is running |
-| `hermes version` | Show version info |
-| `hermes pairing list/approve/revoke` | Manage DM pairing codes |
-
---
-
-## Messaging Gateway
-
-The gateway connects Hermes to Telegram, Discord, Slack, and WhatsApp.
-
-### Setup
-
-The interactive setup wizard handles platform configuration:
-
-```bash
-hermes gateway setup      # Arrow-key menu of all platforms, configure tokens/allowlists/home channels
-```
-
-This is the recommended way to configure messaging. It shows which platforms are already set up, walks through each one interactively, and offers to start/restart the gateway service at the end.
-
-Platforms can also be configured manually in `~/.hermes/.env`:
-
-### Configuration (in `~/.hermes/.env`):
-
-```bash
-# Telegram
-TELEGRAM_BOT_TOKEN=123456:ABC-DEF...      # From @BotFather
-TELEGRAM_ALLOWED_USERS=123456789,987654   # Comma-separated user IDs (from @userinfobot)
-
-# Discord  
-DISCORD_BOT_TOKEN=MTIz...                 # From Developer Portal
-DISCORD_ALLOWED_USERS=123456789012345678  # Comma-separated user IDs
-
-# Agent Behavior
-HERMES_MAX_ITERATIONS=60                  # Max tool-calling iterations
-MESSAGING_CWD=/home/myuser                # Terminal working directory for messaging
-
-# Tool progress is configured in config.yaml (display.tool_progress: off|new|all|verbose)
-```
-
-### Working Directory Behavior
-
- **CLI (`hermes` command)**: Uses current directory (`.` → `os.getcwd()`)
- **Messaging (Telegram/Discord)**: Uses `MESSAGING_CWD` (default: home directory)
-
-This is intentional: CLI users are in a terminal and expect the agent to work in their current directory, while messaging users need a consistent starting location.
-
-### Security (User Allowlists):
-
-**IMPORTANT**: By default, the gateway denies all users who are not in an allowlist or paired via DM.
-
-The gateway checks `{PLATFORM}_ALLOWED_USERS` environment variables:
- If set: Only listed user IDs can interact with the bot
- If unset: All users are denied unless `GATEWAY_ALLOW_ALL_USERS=true` is set
-
-Users can find their IDs:
- **Telegram**: Message [@userinfobot](https://t.me/userinfobot)
- **Discord**: Enable Developer Mode, right-click name → Copy ID
-
-### DM Pairing System
-
-Instead of static allowlists, users can pair via one-time codes:
-1. Unknown user DMs the bot → receives pairing code
-2. Owner runs `hermes pairing approve <platform> <code>`
-3. User is permanently authorized
-
-Security: 8-char codes, 1-hour expiry, rate-limited (1/10min/user), max 3 pending per platform, lockout after 5 failed attempts, `chmod 0600` on data files.
-
-Files: `gateway/pairing.py`, `hermes_cli/pairing.py`
-
-### Event Hooks
-
-Hooks fire at lifecycle points. Place hook directories in `~/.hermes/hooks/`:
-
-```
-~/.hermes/hooks/my-hook/
-├── HOOK.yaml    # name, description, events list
-└── handler.py   # async def handle(event_type, context): ...
-```
-
-Events: `gateway:startup`, `session:start`, `session:reset`, `agent:start`, `agent:step`, `agent:end`, `command:*`
-
-The `agent:step` event fires each iteration of the tool-calling loop with tool names and results.
-
-Files: `gateway/hooks.py`
-
-### Tool Progress Notifications
-
-When `tool_progress` is enabled in `config.yaml`, the bot sends status messages as it works:
- `💻 \`ls -la\`...` (terminal commands show the actual command)
- `🔍 web_search...`
- `📄 web_extract...`
- `🐍 execute_code...` (programmatic tool calling sandbox)
- `🔀 delegate_task...` (subagent delegation)
- `❓ clarify...` (user question, CLI-only)
-
-Modes:
- `new`: Only when switching to a different tool (less spam)
- `all`: Every single tool call
-
-### Typing Indicator
-
-The gateway keeps the "typing..." indicator active throughout processing, refreshing every 4 seconds. This lets users know the bot is working even during long tool-calling sequences.
-
-### Platform Toolsets:
-
-Each platform has a dedicated toolset in `toolsets.py`:
- `hermes-telegram`: Full tools including terminal (with safety checks)
- `hermes-discord`: Full tools including terminal
- `hermes-whatsapp`: Full tools including terminal
-
---
-
-## Configuration System
-
-Configuration files are stored in `~/.hermes/` for easy user access:
- `~/.hermes/config.yaml` - All settings (model, terminal, compression, etc.)
- `~/.hermes/.env` - API keys and secrets
-
-### Adding New Configuration Options
-
-When adding new configuration variables, you MUST follow this process:
-
-#### For config.yaml options:
-
-1. Add to `DEFAULT_CONFIG` in `hermes_cli/config.py`
-2. **CRITICAL**: Bump `_config_version` in `DEFAULT_CONFIG` when adding required fields
-3. This triggers migration prompts for existing users on next `hermes update` or `hermes setup`
-
-Example:
-```python
-DEFAULT_CONFIG = {
-    # ... existing config ...
-    
-    "new_feature": {
-        "enabled": True,
-        "option": "default_value",
-    },
-    
-    # BUMP THIS when adding required fields
-    "_config_version": 2,  # Was 1, now 2
-}
-```
-
-#### For .env variables (API keys/secrets):
-
-1. Add to `REQUIRED_ENV_VARS` or `OPTIONAL_ENV_VARS` in `hermes_cli/config.py`
-2. Include metadata for the migration system:
-
-```python
-OPTIONAL_ENV_VARS = {
-    # ... existing vars ...
-    "NEW_API_KEY": {
-        "description": "What this key is for",
-        "prompt": "Display name in prompts",
-        "url": "https://where-to-get-it.com/",
-        "tools": ["tools_it_enables"],  # What tools need this
-        "password": True,  # Mask input
-    },
-}
-```
-
-#### Update related files:
-
- `hermes_cli/setup.py` - Add prompts in the setup wizard
- `cli-config.yaml.example` - Add example with comments
- Update README.md if user-facing
-
-### Config Version Migration
-
-The system uses `_config_version` to detect outdated configs:
-
-1. `check_for_missing_config()` compares user config to `DEFAULT_CONFIG`
-2. `migrate_config()` interactively prompts for missing values
-3. Called automatically by `hermes update` and optionally by `hermes setup`
-
---
-
-## Environment Variables
-
-API keys are loaded from `~/.hermes/.env`:
- `OPENROUTER_API_KEY` - Main LLM API access (primary provider)
- `FIRECRAWL_API_KEY` - Web search/extract tools
- `FIRECRAWL_API_URL` - Self-hosted Firecrawl endpoint (optional)
- `BROWSERBASE_API_KEY` / `BROWSERBASE_PROJECT_ID` - Browser automation
- `FAL_KEY` - Image generation (FLUX model)
- `NOUS_API_KEY` - Vision and Mixture-of-Agents tools
-
-Terminal tool configuration (in `~/.hermes/config.yaml`):
- `terminal.backend` - Backend: local, docker, singularity, modal, daytona, or ssh
- `terminal.cwd` - Working directory ("." = host CWD for local only; for remote backends set an absolute path inside the target, or omit to use the backend's default)
- `terminal.docker_image` - Image for Docker backend
- `terminal.singularity_image` - Image for Singularity backend
- `terminal.modal_image` - Image for Modal backend
- `terminal.daytona_image` - Image for Daytona backend
- `DAYTONA_API_KEY` - API key for Daytona backend (in .env)
- SSH: `TERMINAL_SSH_HOST`, `TERMINAL_SSH_USER`, `TERMINAL_SSH_KEY` in .env
-
-Agent behavior (in `~/.hermes/.env`):
- `HERMES_MAX_ITERATIONS` - Max tool-calling iterations (default: 60)
- `MESSAGING_CWD` - Working directory for messaging platforms (default: ~)
- `display.tool_progress` in config.yaml - Tool progress: `off`, `new`, `all`, `verbose`
- `OPENAI_API_KEY` - Voice transcription (Whisper STT)
- `SLACK_BOT_TOKEN` / `SLACK_APP_TOKEN` - Slack integration (Socket Mode)
- `SLACK_ALLOWED_USERS` - Comma-separated Slack user IDs
- `HERMES_HUMAN_DELAY_MODE` - Response pacing: off/natural/custom
- `HERMES_HUMAN_DELAY_MIN_MS` / `HERMES_HUMAN_DELAY_MAX_MS` - Custom delay range
-
-### Dangerous Command Approval
-
-The terminal tool includes safety checks for potentially destructive commands (e.g., `rm -rf`, `DROP TABLE`, `chmod 777`, etc.):
-
-**Behavior by Backend:**
- **Docker/Singularity/Modal**: Commands run unrestricted (isolated containers)
- **Local/SSH**: Dangerous commands trigger approval flow
-
-**Approval Flow (CLI):**
-```
-⚠️  Potentially dangerous command detected: recursive delete
-    rm -rf /tmp/test
-
-    [o]nce  |  [s]ession  |  [a]lways  |  [d]eny
-    Choice [o/s/a/D]: 
-```
-
-**Approval Flow (Messaging):**
- Command is blocked with explanation
- Agent explains the command was blocked for safety
- User must add the pattern to their allowlist via `hermes config edit` or run the command directly on their machine
-
-**Configuration:**
- `command_allowlist` in `~/.hermes/config.yaml` stores permanently allowed patterns
- Add patterns via "always" approval or edit directly
-
-**Sudo Handling (Messaging):**
- If sudo fails over messaging, output includes tip to add `SUDO_PASSWORD` to `~/.hermes/.env`
-
---
-
-## Background Process Management
-
-The `process` tool works alongside `terminal` for managing long-running background processes:
-
-**Starting a background process:**
-```python
-terminal(command="pytest -v tests/", background=true)
-# Returns: {"session_id": "proc_abc123", "pid": 12345, ...}
-```
-
-**Managing it with the process tool:**
- `process(action="list")` -- show all running/recent processes
- `process(action="poll", session_id="proc_abc123")` -- check status + new output
- `process(action="log", session_id="proc_abc123")` -- full output with pagination
- `process(action="wait", session_id="proc_abc123", timeout=600)` -- block until done
- `process(action="kill", session_id="proc_abc123")` -- terminate
- `process(action="write", session_id="proc_abc123", data="y")` -- send stdin
- `process(action="submit", session_id="proc_abc123", data="yes")` -- send + Enter
-
-**Key behaviors:**
- Background processes execute through the configured terminal backend (local/Docker/Modal/Daytona/SSH/Singularity) -- never directly on the host unless `TERMINAL_ENV=local`
- The `wait` action blocks the tool call until the process finishes, times out, or is interrupted by a new user message
- PTY mode (`pty=true` on terminal) enables interactive CLI tools (Codex, Claude Code)
- In RL training, background processes are auto-killed when the episode ends (`tool_context.cleanup()`)
- In the gateway, sessions with active background processes are exempt from idle reset
- The process registry checkpoints to `~/.hermes/processes.json` for crash recovery
-
-Files: `tools/process_registry.py` (registry + handler), `tools/terminal_tool.py` (spawn integration)
+1. Add to `COMMANDS` dict in `hermes_cli/commands.py`
+2. Add handler in `HermesCLI.process_command()` in `cli.py`
+3. For persistent settings, use `save_config_value()` in `cli.py`

 ---

 ## Adding New Tools

-Adding a tool requires changes in **2 files** (the tool file and `toolsets.py`):
-
-1. **Create `tools/your_tool.py`** with handler, schema, check function, and registry call:
+Requires changes in **3 files**:

+**1. Create `tools/your_tool.py`:**
 ```python
-# tools/example_tool.py
-import json
-import os
+import json, os
 from tools.registry import registry

-def check_example_requirements() -> bool:
-    """Check if required API keys/dependencies are available."""
+def check_requirements() -> bool:
    return bool(os.getenv("EXAMPLE_API_KEY"))

 def example_tool(param: str, task_id: str = None) -> str:
-    """Execute the tool and return JSON string result."""
-    try:
-        result = {"success": True, "data": "..."}
-        return json.dumps(result, ensure_ascii=False)
-    except Exception as e:
-        return json.dumps({"error": str(e)}, ensure_ascii=False)
-
-EXAMPLE_SCHEMA = {
-    "name": "example_tool",
-    "description": "Does something useful.",
-    "parameters": {
-        "type": "object",
-        "properties": {
-            "param": {"type": "string", "description": "The parameter"}
-        },
-        "required": ["param"]
-    }
-}
+    return json.dumps({"success": True, "data": "..."})

 registry.register(
    name="example_tool",
    toolset="example",
-    schema=EXAMPLE_SCHEMA,
-    handler=lambda args, **kw: example_tool(
-        param=args.get("param", ""), task_id=kw.get("task_id")),
-    check_fn=check_example_requirements,
+    schema={"name": "example_tool", "description": "...", "parameters": {...}},
+    handler=lambda args, **kw: example_tool(param=args.get("param", ""), task_id=kw.get("task_id")),
+    check_fn=check_requirements,
    requires_env=["EXAMPLE_API_KEY"],
 )
 ```

-2. **Add to `toolsets.py`**: Add `"example_tool"` to `_HERMES_CORE_TOOLS` if it should be in all platform toolsets, or create a new toolset entry.
+**2. Add import** in `model_tools.py` `_discover_tools()` list.

-3. **Add discovery import** in `model_tools.py`'s `_discover_tools()` list: `"tools.example_tool"`.
+**3. Add to `toolsets.py`** — either `_HERMES_CORE_TOOLS` (all platforms) or a new toolset.

-That's it. The registry handles schema collection, dispatch, availability checking, and error wrapping automatically. No edits to `TOOLSET_REQUIREMENTS`, `handle_function_call()`, `get_all_tool_names()`, or any other data structure.
+The registry handles schema collection, dispatch, availability checking, and error wrapping. All handlers MUST return a JSON string.

-**Optional:** Add to `OPTIONAL_ENV_VARS` in `hermes_cli/config.py` for the setup wizard, and to `toolset_distributions.py` for batch processing.
-
-**Special case: tools that need agent-level state** (like `todo`, `memory`):
-These are intercepted by `run_agent.py`'s tool dispatch loop *before* `handle_function_call()`. The registry still holds their schemas, but dispatch returns a stub error as a safety fallback. See `todo_tool.py` for the pattern.
-
-All tool handlers MUST return a JSON string. The registry's `dispatch()` wraps all exceptions in `{"error": "..."}` automatically.
-
-### Dynamic Tool Availability
-
-Tools declare their requirements at registration time via `check_fn` and `requires_env`. The registry checks `check_fn()` when building tool definitions -- tools whose check fails are silently excluded.
-
-### Stateful Tools
-
-Tools that maintain state (terminal, browser) require:
- `task_id` parameter for session isolation between concurrent tasks
- `cleanup_*()` function to release resources
- Cleanup is called automatically in run_agent.py after conversation completes
+**Agent-level tools** (todo, memory): intercepted by `run_agent.py` before `handle_function_call()`. See `todo_tool.py` for the pattern.

 ---

-## Trajectory Format
+## Adding Configuration

-Conversations are saved in ShareGPT format for training:
-```json
-{"from": "system", "value": "System prompt with <tools>...</tools>"}
-{"from": "human", "value": "User message"}
-{"from": "gpt", "value": "<think>reasoning</think>\n<tool_call>{...}</tool_call>"}
-{"from": "tool", "value": "<tool_response>{...}</tool_response>"}
-{"from": "gpt", "value": "Final response"}
+### config.yaml options:
+1. Add to `DEFAULT_CONFIG` in `hermes_cli/config.py`
+2. Bump `_config_version` (currently 5) to trigger migration for existing users
+
+### .env variables:
+1. Add to `OPTIONAL_ENV_VARS` in `hermes_cli/config.py` with metadata:
+```python
+"NEW_API_KEY": {
+    "description": "What it's for",
+    "prompt": "Display name",
+    "url": "https://...",
+    "password": True,
+    "category": "tool",  # provider, tool, messaging, setting
+},
 ```

-Tool calls use `<tool_call>` XML tags, responses use `<tool_response>` tags, reasoning uses `<think>` tags.
+### Config loaders (two separate systems):

-### Trajectory Export
+| Loader | Used by | Location |
+|--------|---------|----------|
+| `load_cli_config()` | CLI mode | `cli.py` |
+| `load_config()` | `hermes tools`, `hermes setup` | `hermes_cli/config.py` |
+| Direct YAML load | Gateway | `gateway/run.py` |
+
+---
+
+## Skin/Theme System
+
+The skin engine (`hermes_cli/skin_engine.py`) provides data-driven CLI visual customization. Skins are **pure data** — no code changes needed to add a new skin.
+
+### Architecture
+
+```
+hermes_cli/skin_engine.py    # SkinConfig dataclass, built-in skins, YAML loader
+~/.hermes/skins/*.yaml       # User-installed custom skins (drop-in)
+```
+
+- `init_skin_from_config()` — called at CLI startup, reads `display.skin` from config
+- `get_active_skin()` — returns cached `SkinConfig` for the current skin
+- `set_active_skin(name)` — switches skin at runtime (used by `/skin` command)
+- `load_skin(name)` — loads from user skins first, then built-ins, then falls back to default
+- Missing skin values inherit from the `default` skin automatically
+
+### What skins customize
+
+| Element | Skin Key | Used By |
+|---------|----------|---------|
+| Banner panel border | `colors.banner_border` | `banner.py` |
+| Banner panel title | `colors.banner_title` | `banner.py` |
+| Banner section headers | `colors.banner_accent` | `banner.py` |
+| Banner dim text | `colors.banner_dim` | `banner.py` |
+| Banner body text | `colors.banner_text` | `banner.py` |
+| Response box border | `colors.response_border` | `cli.py` |
+| Spinner faces (waiting) | `spinner.waiting_faces` | `display.py` |
+| Spinner faces (thinking) | `spinner.thinking_faces` | `display.py` |
+| Spinner verbs | `spinner.thinking_verbs` | `display.py` |
+| Spinner wings (optional) | `spinner.wings` | `display.py` |
+| Tool output prefix | `tool_prefix` | `display.py` |
+| Agent name | `branding.agent_name` | `banner.py`, `cli.py` |
+| Welcome message | `branding.welcome` | `cli.py` |
+| Response box label | `branding.response_label` | `cli.py` |
+| Prompt symbol | `branding.prompt_symbol` | `cli.py` |
+
+### Built-in skins
+
+- `default` — Classic Hermes gold/kawaii (the current look)
+- `ares` — Crimson/bronze war-god theme with custom spinner wings
+- `mono` — Clean grayscale monochrome
+- `slate` — Cool blue developer-focused theme
+
+### Adding a built-in skin
+
+Add to `_BUILTIN_SKINS` dict in `hermes_cli/skin_engine.py`:

 ```python
-agent = AIAgent(save_trajectories=True)
-agent.chat("Do something")
-# Saves to trajectories/*.jsonl in ShareGPT format
+"mytheme": {
+    "name": "mytheme",
+    "description": "Short description",
+    "colors": { ... },
+    "spinner": { ... },
+    "branding": { ... },
+    "tool_prefix": "┊",
+},
 ```

---
+### User skins (YAML)

-## Batch Processing (batch_runner.py)
-
-For processing multiple prompts:
- Parallel execution with multiprocessing
- Content-based resume for fault tolerance (matches on prompt text, not indices)
- Toolset distributions control probabilistic tool availability per prompt
- Output: `data/<run_name>/trajectories.jsonl` (combined) + individual batch files
-
-```bash
-python batch_runner.py \
-    --dataset_file=prompts.jsonl \
-    --batch_size=20 \
-    --num_workers=4 \
-    --run_name=my_run
-```
-
---
-
-## Skills System
-
-Skills are on-demand knowledge documents the agent can load. Compatible with the [agentskills.io](https://agentskills.io/specification) open standard.
-
-```
-skills/
-├── mlops/                    # Category folder
-│   ├── axolotl/             # Skill folder
-│   │   ├── SKILL.md         # Main instructions (required)
-│   │   ├── references/      # Additional docs, API specs
-│   │   ├── templates/       # Output formats, configs
-│   │   └── assets/          # Supplementary files (agentskills.io)
-│   └── vllm/
-│       └── SKILL.md
-├── .hub/                    # Skills Hub state (gitignored)
-│   ├── lock.json            # Installed skill provenance
-│   ├── quarantine/          # Pending security review
-│   ├── audit.log            # Security scan history
-│   ├── taps.json            # Custom source repos
-│   └── index-cache/         # Cached remote indexes
-```
-
-**Progressive disclosure** (token-efficient):
-1. `skills_categories()` - List category names (~50 tokens)
-2. `skills_list(category)` - Name + description per skill (~3k tokens)
-3. `skill_view(name)` - Full content + tags + linked files
-
-SKILL.md files use YAML frontmatter (agentskills.io format):
-```yaml
---
-name: skill-name
-description: Brief description for listing
-version: 1.0.0
-platforms: [macos]              # Optional — restrict to specific OS (macos/linux/windows)
-metadata:
-  hermes:
-    tags: [tag1, tag2]
-    related_skills: [other-skill]
---
-# Skill Content...
-```
-
-**Platform filtering** — Skills with a `platforms` field are automatically excluded from the system prompt index, `skills_list()`, and slash commands on incompatible platforms. Skills without the field load everywhere (backward compatible). See `skills/apple/` for macOS-only examples (iMessage, Reminders, Notes, FindMy).
-
-**Skills Hub** — user-driven skill search/install from online registries and official optional skills. Sources: official optional skills (shipped with repo, labeled "official"), GitHub (openai/skills, anthropics/skills, custom taps), ClawHub, Claude marketplace, LobeHub. Not exposed as an agent tool — the model cannot search for or install skills. Users manage skills via `hermes skills browse/search/install` CLI commands or the `/skills` slash command in chat.
-
-Key files:
- `tools/skills_tool.py` — Agent-facing skill list/view (progressive disclosure)
- `tools/skills_guard.py` — Security scanner (regex + LLM audit, trust-aware install policy)
- `tools/skills_hub.py` — Source adapters (OptionalSkillSource, GitHub, ClawHub, Claude marketplace, LobeHub), lock file, auth
- `hermes_cli/skills_hub.py` — CLI subcommands + `/skills` slash command handler
-
---
-
-## Auxiliary Model Configuration
-
-Hermes uses lightweight "auxiliary" models for side tasks that run alongside the main conversation model:
-
-| Task | Tool(s) | Default Model |
-|------|---------|---------------|
-| **Vision analysis** | `vision_analyze`, `browser_vision` | `google/gemini-3-flash-preview` (via OpenRouter) |
-| **Web extraction** | `web_extract`, browser snapshot summarization | `google/gemini-3-flash-preview` (via OpenRouter) |
-| **Context compression** | Auto-compression when approaching context limit | `google/gemini-3-flash-preview` (via OpenRouter) |
-
-By default, these auto-detect the best available provider: OpenRouter → Nous Portal → (text tasks only) custom endpoint → Codex → API-key providers.
-
-### Changing the Vision Model
-
-To use a different model for image analysis (e.g., GPT-4o instead of Gemini Flash), add to `~/.hermes/config.yaml`:
+Users create `~/.hermes/skins/<name>.yaml`:

 ```yaml
-auxiliary:
-  vision:
-    provider: "openrouter"        # or "nous", "main", "auto"
-    model: "openai/gpt-4o"        # any model slug your provider supports
+name: cyberpunk
+description: Neon-soaked terminal theme
+
+colors:
+  banner_border: "#FF00FF"
+  banner_title: "#00FFFF"
+  banner_accent: "#FF1493"
+
+spinner:
+  thinking_verbs: ["jacking in", "decrypting", "uploading"]
+  wings:
+    - ["⟨⚡", "⚡⟩"]
+
+branding:
+  agent_name: "Cyber Agent"
+  response_label: " ⚡ Cyber "
+
+tool_prefix: "▏"
 ```

-Or set environment variables (in `~/.hermes/.env` or shell):
+Activate with `/skin cyberpunk` or `display.skin: cyberpunk` in config.yaml.

-```bash
-AUXILIARY_VISION_MODEL=openai/gpt-4o
-# Optionally force a specific provider:
-AUXILIARY_VISION_PROVIDER=openrouter
-```
+---

-### Changing the Web Extraction Model
+## Important Policies

-```yaml
-auxiliary:
-  web_extract:
-    provider: "auto"
-    model: "google/gemini-2.5-flash"
-```
+### Prompt Caching Must Not Break

-### Changing the Compression Model
+Hermes-Agent ensures caching remains valid throughout a conversation. **Do NOT implement changes that would:**
+- Alter past context mid-conversation
+- Change toolsets mid-conversation
+- Reload memories or rebuild system prompts mid-conversation

-```yaml
-compression:
-  summary_model: "google/gemini-2.5-flash"
-  summary_provider: "auto"          # "auto", "openrouter", "nous", "main"
-```
+Cache-breaking forces dramatically higher costs. The ONLY time we alter context is during context compression.

-### Provider Options
+### Working Directory Behavior
+- **CLI**: Uses current directory (`.` → `os.getcwd()`)
+- **Messaging**: Uses `MESSAGING_CWD` env var (default: home directory)

-| Provider | Description |
-|----------|-------------|
-| `"auto"` | Best available (default). For vision, only tries OpenRouter + Nous. |
-| `"openrouter"` | Force OpenRouter (requires `OPENROUTER_API_KEY`) |
-| `"nous"` | Force Nous Portal (requires `hermes login`) |
-| `"codex"` | Force Codex OAuth (ChatGPT account). Supports vision via gpt-5.3-codex. |
-| `"main"` | Use your custom endpoint (`OPENAI_BASE_URL` + `OPENAI_API_KEY`). Works with OpenAI API, local models, etc. |
+### Background Process Notifications (Gateway)

-**Important:** Vision tasks require a multimodal-capable model. In `auto` mode, OpenRouter, Nous Portal, and Codex OAuth are tried (they all support vision). Setting `provider: "main"` for vision will work only if your endpoint supports multimodal input (e.g. OpenAI with GPT-4o, or a local model with vision).
+When `terminal(background=true, check_interval=...)` is used, the gateway runs a watcher that
+pushes status updates to the user's chat. Control verbosity with `display.background_process_notifications`
+in config.yaml (or `HERMES_BACKGROUND_NOTIFICATIONS` env var):

-**Key files:** `agent/auxiliary_client.py` (resolution chain), `tools/vision_tools.py`, `tools/browser_tool.py`, `tools/web_tools.py`
+- `all` — running-output updates + final message (default)
+- `result` — only the final completion message
+- `error` — only the final message when exit code != 0
+- `off` — no watcher messages at all

 ---

 ## Known Pitfalls

 ### DO NOT use `simple_term_menu` for interactive menus
-
-`simple_term_menu` has rendering bugs in tmux, iTerm2, and other non-standard terminals. When the user scrolls with arrow keys, previously highlighted items "ghost" — duplicating upward and corrupting the display. This happens because the library uses ANSI cursor-up codes to redraw in place, and tmux/iTerm miscalculate positions when the menu is near the bottom of the viewport.
-
-**Rule:** All interactive menus in `hermes_cli/` must use `curses` (Python stdlib) instead. See `tools_config.py` for the pattern — both `_prompt_choice()` (single-select) and `_prompt_toolset_checklist()` (multi-select with space toggle) use `curses.wrapper()`. The numbered-input fallback handles Windows where curses isn't available.
+Rendering bugs in tmux/iTerm2 — ghosting on scroll. Use `curses` (stdlib) instead. See `hermes_cli/tools_config.py` for the pattern.

 ### DO NOT use `\033[K` (ANSI erase-to-EOL) in spinner/display code
-
-The ANSI escape `\033[K` leaks as literal `?[K` text when `prompt_toolkit`'s `patch_stdout` is active. Use space-padding instead to clear lines: `f"\r{line}{' ' * pad}"`. See `agent/display.py` `KawaiiSpinner`.
+Leaks as literal `?[K` text under `prompt_toolkit`'s `patch_stdout`. Use space-padding: `f"\r{line}{' ' * pad}"`.

 ### `_last_resolved_tool_names` is a process-global in `model_tools.py`
-
-The `execute_code` sandbox uses `_last_resolved_tool_names` (set by `get_tool_definitions()`) to decide which tool stubs to generate. When subagents run with restricted toolsets, they overwrite this global. After delegation returns to the parent, `execute_code` may see the child's restricted list instead of the parent's full list. This is a known bug — `execute_code` calls after delegation may fail with `ImportError: cannot import name 'patch' from 'hermes_tools'`.
+When subagents overwrite this global, `execute_code` calls after delegation may fail with missing tool imports. Known bug.

 ### Tests must not write to `~/.hermes/`
-
-The `autouse` fixture `_isolate_hermes_home` in `tests/conftest.py` redirects `HERMES_HOME` to a temp dir. Every test runs in isolation. If you add a test that creates `AIAgent` instances or writes session logs, the fixture handles cleanup automatically. Never hardcode `~/.hermes/` paths in tests.
+The `_isolate_hermes_home` autouse fixture in `tests/conftest.py` redirects `HERMES_HOME` to a temp dir. Never hardcode `~/.hermes/` paths in tests.

 ---

-## Testing Changes
+## Testing

-After making changes:
+```bash
+source .venv/bin/activate
+python -m pytest tests/ -q          # Full suite (~2500 tests, ~2 min)
+python -m pytest tests/test_model_tools.py -q   # Toolset resolution
+python -m pytest tests/test_cli_init.py -q       # CLI config loading
+python -m pytest tests/gateway/ -q               # Gateway tests
+python -m pytest tests/tools/ -q                 # Tool-level tests
+```

-1. Run `hermes doctor` to check setup
-2. Run `hermes config check` to verify config
-3. Test with `hermes chat -q "test message"`
-4. For new config options, test fresh install: `rm -rf ~/.hermes && hermes setup`
+Always run the full suite before pushing changes.
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -139,7 +139,8 @@ hermes-agent/
 │   ├── commands.py               # Slash command definitions + autocomplete
 │   ├── callbacks.py              # Interactive callbacks (clarify, sudo, approval)
 │   ├── doctor.py                 # Diagnostics
-│   └── skills_hub.py             # Skills Hub CLI + /skills slash command
+│   ├── skills_hub.py             # Skills Hub CLI + /skills slash command
+│   └── skin_engine.py            # Skin/theme engine — data-driven CLI visual customization
 │
 ├── tools/                    # Tool implementations (self-registering)
 │   ├── registry.py               # Central tool registry (schemas, handlers, dispatch)
@@ -375,6 +376,56 @@ If the field is omitted or empty, the skill loads on all platforms (backward com

 ---

+## Adding a Skin / Theme
+
+Hermes uses a data-driven skin system — no code changes needed to add a new skin.
+
+**Option A: User skin (YAML file)**
+
+Create `~/.hermes/skins/<name>.yaml`:
+
+```yaml
+name: mytheme
+description: Short description of the theme
+
+colors:
+  banner_border: "#HEX"     # Panel border color
+  banner_title: "#HEX"      # Panel title color
+  banner_accent: "#HEX"     # Section header color
+  banner_dim: "#HEX"        # Muted/dim text color
+  banner_text: "#HEX"       # Body text color
+  response_border: "#HEX"   # Response box border
+
+spinner:
+  waiting_faces: ["(⚔)", "(⛨)"]
+  thinking_faces: ["(⚔)", "(⌁)"]
+  thinking_verbs: ["forging", "plotting"]
+  wings:                     # Optional left/right decorations
+    - ["⟪⚔", "⚔⟫"]
+
+branding:
+  agent_name: "My Agent"
+  welcome: "Welcome message"
+  response_label: " ⚔ Agent "
+  prompt_symbol: "⚔ ❯ "
+
+tool_prefix: "╎"             # Tool output line prefix
+```
+
+All fields are optional — missing values inherit from the default skin.
+
+**Option B: Built-in skin**
+
+Add to `_BUILTIN_SKINS` dict in `hermes_cli/skin_engine.py`. Use the same schema as above but as a Python dict. Built-in skins ship with the package and are always available.
+
+**Activating:**
+- CLI: `/skin mytheme` or set `display.skin: mytheme` in config.yaml
+- Config: `display: { skin: mytheme }`
+
+See `hermes_cli/skin_engine.py` for the full schema and existing skins as examples.
+
+---
+
 ## Cross-Platform Compatibility

 Hermes runs on Linux, macOS, and Windows. When writing code that touches the OS:
--- a/README.md
+++ b/README.md
@@ -17,7 +17,7 @@ Use any model you want — [Nous Portal](https://portal.nousresearch.com), [Open

 <table>
 <tr><td><b>A real terminal interface</b></td><td>Full TUI with multiline editing, slash-command autocomplete, conversation history, interrupt-and-redirect, and streaming tool output.</td></tr>
-<tr><td><b>Lives where you do</b></td><td>Telegram, Discord, Slack, WhatsApp, and CLI — all from a single gateway process. Voice memo transcription, cross-platform conversation continuity.</td></tr>
+<tr><td><b>Lives where you do</b></td><td>Telegram, Discord, Slack, WhatsApp, Signal, and CLI — all from a single gateway process. Voice memo transcription, cross-platform conversation continuity.</td></tr>
 <tr><td><b>A closed learning loop</b></td><td>Agent-curated memory with periodic nudges. Autonomous skill creation after complex tasks. Skills self-improve during use. FTS5 session search with LLM summarization for cross-session recall. <a href="https://github.com/plastic-labs/honcho">Honcho</a> dialectic user modeling. Compatible with the <a href="https://agentskills.io">agentskills.io</a> open standard.</td></tr>
 <tr><td><b>Scheduled automations</b></td><td>Built-in cron scheduler with delivery to any platform. Daily reports, nightly backups, weekly audits — all in natural language, running unattended.</td></tr>
 <tr><td><b>Delegates and parallelizes</b></td><td>Spawn isolated subagents for parallel workstreams. Write Python scripts that call tools via RPC, collapsing multi-step pipelines into zero-context-cost turns.</td></tr>
@@ -71,7 +71,7 @@ All documentation lives at **[hermes-agent.nousresearch.com/docs](https://hermes
 | [Quickstart](https://hermes-agent.nousresearch.com/docs/getting-started/quickstart) | Install → setup → first conversation in 2 minutes |
 | [CLI Usage](https://hermes-agent.nousresearch.com/docs/user-guide/cli) | Commands, keybindings, personalities, sessions |
 | [Configuration](https://hermes-agent.nousresearch.com/docs/user-guide/configuration) | Config file, providers, models, all options |
-| [Messaging Gateway](https://hermes-agent.nousresearch.com/docs/user-guide/messaging) | Telegram, Discord, Slack, WhatsApp, Home Assistant |
+| [Messaging Gateway](https://hermes-agent.nousresearch.com/docs/user-guide/messaging) | Telegram, Discord, Slack, WhatsApp, Signal, Home Assistant |
 | [Security](https://hermes-agent.nousresearch.com/docs/user-guide/security) | Command approval, DM pairing, container isolation |
 | [Tools & Toolsets](https://hermes-agent.nousresearch.com/docs/user-guide/features/tools) | 40+ tools, toolset system, terminal backends |
 | [Skills System](https://hermes-agent.nousresearch.com/docs/user-guide/features/skills) | Procedural memory, Skills Hub, creating skills |
--- a/agent/auxiliary_client.py
+++ b/agent/auxiliary_client.py
@@ -560,12 +560,16 @@ def get_vision_auxiliary_client() -> Tuple[Optional[OpenAI], Optional[str]]:
    forced = _get_auxiliary_provider("vision")
    if forced != "auto":
        return _resolve_forced_provider(forced)
-    # Auto: only multimodal-capable providers
-    for try_fn in (_try_openrouter, _try_nous, _try_codex):
+    # Auto: try providers known to support multimodal first, then fall
+    # back to the user's custom endpoint.  Many local models (Qwen-VL,
+    # LLaVA, Pixtral, etc.) support vision — skipping them entirely
+    # caused silent failures for local-only users.
+    for try_fn in (_try_openrouter, _try_nous, _try_codex,
+                   _try_custom_endpoint):
        client, model = try_fn()
        if client is not None:
            return client, model
-    logger.debug("Auxiliary vision client: none available (auto only tries OpenRouter/Nous/Codex)")
+    logger.debug("Auxiliary vision client: none available")
    return None, None


--- a/agent/context_compressor.py
+++ b/agent/context_compressor.py
@@ -342,7 +342,9 @@ Write only the summary, starting with "[CONTEXT SUMMARY]:" prefix."""
            compressed.append(msg)

        if summary:
-            compressed.append({"role": "user", "content": summary})
+            last_head_role = messages[compress_start - 1].get("role", "user") if compress_start > 0 else "user"
+            summary_role = "user" if last_head_role in ("assistant", "tool") else "assistant"
+            compressed.append({"role": summary_role, "content": summary})
        else:
            if not self.quiet_mode:
                print("   ⚠️  No summary model available — middle turns dropped without summary")
--- a/agent/display.py
+++ b/agent/display.py
@@ -5,8 +5,8 @@ Used by AIAgent._execute_tool_calls for CLI feedback.
 """

 import json
+import logging
 import os
-import random
 import sys
 import threading
 import time
@@ -15,6 +15,49 @@ import time
 _RED = "\033[31m"
 _RESET = "\033[0m"

+logger = logging.getLogger(__name__)
+
+
+# =========================================================================
+# Skin-aware helpers (lazy import to avoid circular deps)
+# =========================================================================
+
+def _get_skin():
+    """Get the active skin config, or None if not available."""
+    try:
+        from hermes_cli.skin_engine import get_active_skin
+        return get_active_skin()
+    except Exception:
+        return None
+
+
+def get_skin_faces(key: str, default: list) -> list:
+    """Get spinner face list from active skin, falling back to default."""
+    skin = _get_skin()
+    if skin:
+        faces = skin.get_spinner_list(key)
+        if faces:
+            return faces
+    return default
+
+
+def get_skin_verbs() -> list:
+    """Get thinking verbs from active skin."""
+    skin = _get_skin()
+    if skin:
+        verbs = skin.get_spinner_list("thinking_verbs")
+        if verbs:
+            return verbs
+    return KawaiiSpinner.THINKING_VERBS
+
+
+def get_skin_tool_prefix() -> str:
+    """Get tool output prefix character from active skin."""
+    skin = _get_skin()
+    if skin:
+        return skin.tool_prefix
+    return "┊"
+

 # =========================================================================
 # Tool preview (one-line summary of a tool call's primary argument)
@@ -22,6 +65,8 @@ _RESET = "\033[0m"

 def build_tool_preview(tool_name: str, args: dict, max_len: int = 40) -> str:
    """Build a short preview of a tool call's primary argument for display."""
+    if not args:
+        return None
    primary_args = {
        "terminal": "command", "web_search": "query", "web_extract": "urls",
        "read_file": "path", "write_file": "path", "patch": "path",
@@ -163,6 +208,7 @@ class KawaiiSpinner:
        self.frame_idx = 0
        self.start_time = None
        self.last_line_len = 0
+        self._last_flush_time = 0.0  # Rate-limit flushes for patch_stdout compat
        # Capture stdout NOW, before any redirect_stdout(devnull) from
        # child agents can replace sys.stdout with a black hole.
        self._out = sys.stdout
@@ -177,15 +223,34 @@ class KawaiiSpinner:
            pass

    def _animate(self):
+        # Cache skin wings at start (avoid per-frame imports)
+        skin = _get_skin()
+        wings = skin.get_spinner_wings() if skin else []
+
        while self.running:
            if os.getenv("HERMES_SPINNER_PAUSE"):
                time.sleep(0.1)
                continue
            frame = self.spinner_frames[self.frame_idx % len(self.spinner_frames)]
            elapsed = time.time() - self.start_time
-            line = f"  {frame} {self.message} ({elapsed:.1f}s)"
+            if wings:
+                left, right = wings[self.frame_idx % len(wings)]
+                line = f"  {left} {frame} {self.message} {right} ({elapsed:.1f}s)"
+            else:
+                line = f"  {frame} {self.message} ({elapsed:.1f}s)"
            pad = max(self.last_line_len - len(line), 0)
-            self._write(f"\r{line}{' ' * pad}", end='', flush=True)
+            # Rate-limit flush() calls to avoid spinner spam under
+            # prompt_toolkit's patch_stdout.  Each flush() pushes a queue
+            # item that may trigger a separate run_in_terminal() call; if
+            # items are processed one-at-a-time the \r overwrite is lost
+            # and every frame appears on its own line.  By flushing at
+            # most every 0.4s we guarantee multiple \r-frames are batched
+            # into a single write, so the terminal collapses them correctly.
+            now = time.time()
+            should_flush = (now - self._last_flush_time) >= 0.4
+            self._write(f"\r{line}{' ' * pad}", end='', flush=should_flush)
+            if should_flush:
+                self._last_flush_time = now
            self.last_line_len = len(line)
            self.frame_idx += 1
            time.sleep(0.12)
@@ -300,7 +365,7 @@ def _detect_tool_failure(tool_name: str, result: str | None) -> tuple[bool, str]
            if exit_code is not None and exit_code != 0:
                return True, f" [exit {exit_code}]"
        except (json.JSONDecodeError, TypeError, AttributeError):
-            pass
+            logger.debug("Could not parse terminal result as JSON for exit code check")
        return False, ""

    # Memory-specific: distinguish "full" from real errors
@@ -310,7 +375,7 @@ def _detect_tool_failure(tool_name: str, result: str | None) -> tuple[bool, str]
            if data.get("success") is False and "exceed the limit" in data.get("error", ""):
                return True, " [full]"
        except (json.JSONDecodeError, TypeError, AttributeError):
-            pass
+            logger.debug("Could not parse memory result as JSON for capacity check")

    # Generic heuristic for non-terminal tools
    lower = result[:500].lower()
@@ -332,6 +397,7 @@ def get_cute_tool_message(
    """
    dur = f"{duration:.1f}s"
    is_failure, failure_suffix = _detect_tool_failure(tool_name, result)
+    skin_prefix = get_skin_tool_prefix()

    def _trunc(s, n=40):
        s = str(s)
@@ -342,7 +408,9 @@ def get_cute_tool_message(
        return ("..." + p[-(n-3):]) if len(p) > n else p

    def _wrap(line: str) -> str:
-        """Append failure suffix when the tool failed."""
+        """Apply skin tool prefix and failure suffix."""
+        if skin_prefix != "┊":
+            line = line.replace("┊", skin_prefix, 1)
        if not is_failure:
            return line
        return f"{line}{failure_suffix}"
--- a/agent/prompt_builder.py
+++ b/agent/prompt_builder.py
@@ -122,6 +122,15 @@ PLATFORM_HINTS = {
        "attachments, audio as file attachments. You can also include image URLs "
        "in markdown format ![alt](url) and they will be uploaded as attachments."
    ),
+    "signal": (
+        "You are on a text messaging communication platform, Signal. "
+        "Please do not use markdown as it does not render. "
+        "You can send media files natively: to deliver a file to the user, "
+        "include MEDIA:/absolute/path/to/file in your response. Images "
+        "(.png, .jpg, .webp) appear as photos, audio as attachments, and other "
+        "files arrive as downloadable documents. You can also include image "
+        "URLs in markdown format ![alt](url) and they will be sent as photos."
+    ),
    "cli": (
        "You are a CLI AI Agent. Try not to use markdown but simple text "
        "renderable inside a terminal."
@@ -150,8 +159,8 @@ def _read_skill_description(skill_file: Path, max_chars: int = 60) -> str:
            if len(desc) > max_chars:
                desc = desc[:max_chars - 3] + "..."
            return desc
-    except Exception:
-        pass
+    except Exception as e:
+        logger.debug("Failed to read skill description from %s: %s", skill_file, e)
    return ""


@@ -186,6 +195,8 @@ def build_skills_system_prompt() -> str:

    # Collect skills with descriptions, grouped by category
    # Each entry: (skill_name, description)
+    # Supports sub-categories: skills/mlops/training/axolotl/SKILL.md
+    # → category "mlops/training", skill "axolotl"
    skills_by_category: dict[str, list[tuple[str, str]]] = {}
    for skill_file in skills_dir.rglob("SKILL.md"):
        # Skip skills incompatible with the current OS platform
@@ -194,8 +205,13 @@ def build_skills_system_prompt() -> str:
        rel_path = skill_file.relative_to(skills_dir)
        parts = rel_path.parts
        if len(parts) >= 2:
-            category = parts[0]
+            # Category is everything between skills_dir and the skill folder
+            # e.g. parts = ("mlops", "training", "axolotl", "SKILL.md")
+            #   → category = "mlops/training", skill_name = "axolotl"
+            # e.g. parts = ("github", "github-auth", "SKILL.md")
+            #   → category = "github", skill_name = "github-auth"
            skill_name = parts[-2]
+            category = "/".join(parts[:-2]) if len(parts) > 2 else parts[0]
        else:
            category = "general"
            skill_name = skill_file.parent.name
@@ -206,9 +222,11 @@ def build_skills_system_prompt() -> str:
        return ""

    # Read category-level descriptions from DESCRIPTION.md
+    # Checks both the exact category path and parent directories
    category_descriptions = {}
    for category in skills_by_category:
-        desc_file = skills_dir / category / "DESCRIPTION.md"
+        cat_path = Path(category)
+        desc_file = skills_dir / cat_path / "DESCRIPTION.md"
        if desc_file.exists():
            try:
                content = desc_file.read_text(encoding="utf-8")
--- a/agent/redact.py
+++ b/agent/redact.py
@@ -8,14 +8,14 @@ the first 6 and last 4 characters for debuggability.
 """

 import logging
+import os
 import re
-from typing import Optional

 logger = logging.getLogger(__name__)

 # Known API key prefixes -- match the prefix + contiguous token chars
 _PREFIX_PATTERNS = [
-    r"sk-[A-Za-z0-9_-]{10,}",           # OpenAI / OpenRouter
+    r"sk-[A-Za-z0-9_-]{10,}",           # OpenAI / OpenRouter / Anthropic (sk-ant-*)
    r"ghp_[A-Za-z0-9]{10,}",            # GitHub PAT (classic)
    r"github_pat_[A-Za-z0-9_]{10,}",    # GitHub PAT (fine-grained)
    r"xox[baprs]-[A-Za-z0-9-]{10,}",    # Slack tokens
@@ -25,6 +25,18 @@ _PREFIX_PATTERNS = [
    r"fc-[A-Za-z0-9]{10,}",             # Firecrawl
    r"bb_live_[A-Za-z0-9_-]{10,}",      # BrowserBase
    r"gAAAA[A-Za-z0-9_=-]{20,}",        # Codex encrypted tokens
+    r"AKIA[A-Z0-9]{16}",                # AWS Access Key ID
+    r"sk_live_[A-Za-z0-9]{10,}",        # Stripe secret key (live)
+    r"sk_test_[A-Za-z0-9]{10,}",        # Stripe secret key (test)
+    r"rk_live_[A-Za-z0-9]{10,}",        # Stripe restricted key
+    r"SG\.[A-Za-z0-9_-]{10,}",          # SendGrid API key
+    r"hf_[A-Za-z0-9]{10,}",             # HuggingFace token
+    r"r8_[A-Za-z0-9]{10,}",             # Replicate API token
+    r"npm_[A-Za-z0-9]{10,}",            # npm access token
+    r"pypi-[A-Za-z0-9_-]{10,}",         # PyPI API token
+    r"dop_v1_[A-Za-z0-9]{10,}",         # DigitalOcean PAT
+    r"doo_v1_[A-Za-z0-9]{10,}",         # DigitalOcean OAuth
+    r"am_[A-Za-z0-9_-]{10,}",           # AgentMail API key
 ]

 # ENV assignment patterns: KEY=value where KEY contains a secret-like name
@@ -52,6 +64,22 @@ _TELEGRAM_RE = re.compile(
    r"(bot)?(\d{8,}):([-A-Za-z0-9_]{30,})",
 )

+# Private key blocks: -----BEGIN RSA PRIVATE KEY----- ... -----END RSA PRIVATE KEY-----
+_PRIVATE_KEY_RE = re.compile(
+    r"-----BEGIN[A-Z ]*PRIVATE KEY-----[\s\S]*?-----END[A-Z ]*PRIVATE KEY-----"
+)
+
+# Database connection strings: protocol://user:PASSWORD@host
+# Catches postgres, mysql, mongodb, redis, amqp URLs and redacts the password
+_DB_CONNSTR_RE = re.compile(
+    r"((?:postgres(?:ql)?|mysql|mongodb(?:\+srv)?|redis|amqp)://[^:]+:)([^@]+)(@)",
+    re.IGNORECASE,
+)
+
+# E.164 phone numbers: +<country><number>, 7-15 digits
+# Negative lookahead prevents matching hex strings or identifiers
+_SIGNAL_PHONE_RE = re.compile(r"(\+[1-9]\d{6,14})(?![A-Za-z0-9])")
+
 # Compile known prefix patterns into one alternation
 _PREFIX_RE = re.compile(
    r"(?<![A-Za-z0-9_-])(" + "|".join(_PREFIX_PATTERNS) + r")(?![A-Za-z0-9_-])"
@@ -69,9 +97,12 @@ def redact_sensitive_text(text: str) -> str:
    """Apply all redaction patterns to a block of text.

    Safe to call on any string -- non-matching text passes through unchanged.
+    Disabled when security.redact_secrets is false in config.yaml.
    """
    if not text:
        return text
+    if os.getenv("HERMES_REDACT_SECRETS", "").lower() in ("0", "false", "no", "off"):
+        return text

    # Known prefixes (sk-, ghp_, etc.)
    text = _PREFIX_RE.sub(lambda m: _mask_token(m.group(1)), text)
@@ -101,6 +132,20 @@ def redact_sensitive_text(text: str) -> str:
        return f"{prefix}{digits}:***"
    text = _TELEGRAM_RE.sub(_redact_telegram, text)

+    # Private key blocks
+    text = _PRIVATE_KEY_RE.sub("[REDACTED PRIVATE KEY]", text)
+
+    # Database connection string passwords
+    text = _DB_CONNSTR_RE.sub(lambda m: f"{m.group(1)}***{m.group(3)}", text)
+
+    # E.164 phone numbers (Signal, WhatsApp)
+    def _redact_phone(m):
+        phone = m.group(1)
+        if len(phone) <= 8:
+            return phone[:2] + "****" + phone[-2:]
+        return phone[:4] + "****" + phone[-4:]
+    text = _SIGNAL_PHONE_RE.sub(_redact_phone, text)
+
    return text


--- a/batch_runner.py
+++ b/batch_runner.py
@@ -606,7 +606,7 @@ class BatchRunner:
        # Create batches
        self.batches = self._create_batches()
        
-        print(f"📊 Batch Runner Initialized")
+        print("📊 Batch Runner Initialized")
        print(f"   Dataset: {self.dataset_file} ({len(self.dataset)} prompts)")
        print(f"   Batch size: {self.batch_size}")
        print(f"   Total batches: {len(self.batches)}")
@@ -826,7 +826,7 @@ class BatchRunner:
            print("=" * 70)
            print(f"   Original dataset size:     {len(self.dataset):,} prompts")
            print(f"   Already completed:         {len(skipped_indices):,} prompts")
-            print(f"   ─────────────────────────────────────────")
+            print("   ─────────────────────────────────────────")
            print(f"   🎯 RESUMING WITH:          {len(filtered_entries):,} prompts")
            print(f"   New batches created:       {len(batches_to_process)}")
            print("=" * 70 + "\n")
@@ -888,7 +888,7 @@ class BatchRunner:
            ]
            
            print(f"✅ Created {len(tasks)} batch tasks")
-            print(f"🚀 Starting parallel batch processing...\n")
+            print("🚀 Starting parallel batch processing...\n")
            
            # Use rich Progress for better visual tracking with persistent bottom bar
            # redirect_stdout/stderr lets rich manage all output so progress bar stays clean
@@ -1057,7 +1057,7 @@ class BatchRunner:
        print(f"✅ Total trajectories in merged file: {total_entries - filtered_entries}")
        print(f"✅ Total batch files merged: {batch_files_found}")
        print(f"⏱️  Total duration: {round(time.time() - start_time, 2)}s")
-        print(f"\n📈 Tool Usage Statistics:")
+        print("\n📈 Tool Usage Statistics:")
        print("-" * 70)
        
        if total_tool_stats:
@@ -1084,7 +1084,7 @@ class BatchRunner:
        # Print reasoning coverage stats
        total_discarded = sum(r.get("discarded_no_reasoning", 0) for r in results)
        
-        print(f"\n🧠 Reasoning Coverage:")
+        print("\n🧠 Reasoning Coverage:")
        print("-" * 70)
        total_turns = total_reasoning_stats["total_assistant_turns"]
        with_reasoning = total_reasoning_stats["turns_with_reasoning"]
@@ -1101,8 +1101,8 @@ class BatchRunner:
            print(f"   🚫 Samples discarded (zero reasoning): {total_discarded:,}")
        
        print(f"\n💾 Results saved to: {self.output_dir}")
-        print(f"   - Trajectories: trajectories.jsonl (combined)")
-        print(f"   - Individual batches: batch_*.jsonl (for debugging)")
+        print("   - Trajectories: trajectories.jsonl (combined)")
+        print("   - Individual batches: batch_*.jsonl (for debugging)")
        print(f"   - Statistics: {self.stats_file.name}")
        print(f"   - Checkpoint: {self.checkpoint_file.name}")

@@ -1238,7 +1238,7 @@ def main(
            with open(prefill_messages_file, 'r', encoding='utf-8') as f:
                prefill_messages = json.load(f)
            if not isinstance(prefill_messages, list):
-                print(f"❌ Error: prefill_messages_file must contain a JSON array of messages")
+                print("❌ Error: prefill_messages_file must contain a JSON array of messages")
                return
            print(f"💬 Loaded {len(prefill_messages)} prefill messages from {prefill_messages_file}")
        except Exception as e:
--- a/cli-config.yaml.example
+++ b/cli-config.yaml.example
@@ -11,6 +11,7 @@ model:
  
  # Inference provider selection:
  #   "auto"       - Use Nous Portal if logged in, otherwise OpenRouter/env vars (default)
+  #   "nous-api"   - Use Nous Portal via API key (requires: NOUS_API_KEY)
  #   "openrouter" - Always use OpenRouter API key from OPENROUTER_API_KEY
  #   "nous"       - Always use Nous Portal (requires: hermes login)
  #   "zai"        - Use z.ai / ZhipuAI GLM models (requires: GLM_API_KEY)
@@ -402,11 +403,13 @@ agent:
 #     discord: [web, vision, skills, todo]
 #
 # If not set, defaults are:
-#   cli:      hermes-cli      (everything + cronjob management)
-#   telegram: hermes-telegram  (terminal, file, web, vision, image, tts, browser, skills, todo, cronjob, messaging)
-#   discord:  hermes-discord   (same as telegram)
-#   whatsapp: hermes-whatsapp  (same as telegram)
-#   slack:    hermes-slack     (same as telegram)
+#   cli:           hermes-cli            (everything + cronjob management)
+#   telegram:      hermes-telegram       (terminal, file, web, vision, image, tts, browser, skills, todo, cronjob, messaging)
+#   discord:       hermes-discord        (same as telegram)
+#   whatsapp:      hermes-whatsapp       (same as telegram)
+#   slack:         hermes-slack          (same as telegram)
+#   signal:        hermes-signal         (same as telegram)
+#   homeassistant: hermes-homeassistant  (same as telegram)
 #
 platform_toolsets:
  cli: [hermes-cli]
@@ -414,6 +417,8 @@ platform_toolsets:
  discord: [hermes-discord]
  whatsapp: [hermes-whatsapp]
  slack: [hermes-slack]
+  signal: [hermes-signal]
+  homeassistant: [hermes-homeassistant]

 # ─────────────────────────────────────────────────────────────────────────────
 # Available toolsets (use these names in platform_toolsets or the toolsets list)
@@ -555,6 +560,21 @@ toolsets:
 #     args: ["-y", "@modelcontextprotocol/server-github"]
 #     env:
 #       GITHUB_PERSONAL_ACCESS_TOKEN: "ghp_..."
+#
+# Sampling (server-initiated LLM requests) — enabled by default.
+# Per-server config under the 'sampling' key:
+#   analysis:
+#     command: npx
+#     args: ["-y", "analysis-server"]
+#     sampling:
+#       enabled: true           # default: true
+#       model: "gemini-3-flash" # override model (optional)
+#       max_tokens_cap: 4096    # max tokens per request
+#       timeout: 30             # LLM call timeout (seconds)
+#       max_rpm: 10             # max requests per minute
+#       allowed_models: []      # model whitelist (empty = all)
+#       max_tool_rounds: 5      # tool loop limit (0 = disable)
+#       log_level: "info"       # audit verbosity

 # =============================================================================
 # Voice Transcription (Speech-to-Text)
@@ -636,7 +656,57 @@ display:
  # Toggle at runtime with /verbose in the CLI
  tool_progress: all

+  # Background process notifications (gateway/messaging only).
+  # Controls how chatty the process watcher is when you use
+  # terminal(background=true, check_interval=...) from Telegram/Discord/etc.
+  #   off:     No watcher messages at all
+  #   result:  Only the final completion message
+  #   error:   Only the final message when exit code != 0
+  #   all:     Running output updates + final message (default)
+  background_process_notifications: all
+
  # Play terminal bell when agent finishes a response.
  # Useful for long-running tasks — your terminal will ding when the agent is done.
  # Works over SSH. Most terminals can be configured to flash the taskbar or play a sound.
  bell_on_complete: false
+
+  # ───────────────────────────────────────────────────────────────────────────
+  # Skin / Theme
+  # ───────────────────────────────────────────────────────────────────────────
+  # Customize CLI visual appearance — banner colors, spinner faces, tool prefix,
+  # response box label, and branding text. Change at runtime with /skin <name>.
+  #
+  # Built-in skins:
+  #   default  — Classic Hermes gold/kawaii
+  #   ares     — Crimson/bronze war-god theme with spinner wings
+  #   mono     — Clean grayscale monochrome
+  #   slate    — Cool blue developer-focused
+  #
+  # Custom skins: drop a YAML file in ~/.hermes/skins/<name>.yaml
+  # Schema (all fields optional, missing values inherit from default):
+  #
+  #   name: my-theme
+  #   description: Short description
+  #   colors:
+  #     banner_border: "#HEX"    # Panel border
+  #     banner_title: "#HEX"     # Panel title
+  #     banner_accent: "#HEX"    # Section headers (Available Tools, etc.)
+  #     banner_dim: "#HEX"       # Dim/muted text
+  #     banner_text: "#HEX"      # Body text (tool names, skill names)
+  #     ui_accent: "#HEX"        # UI accent color
+  #     response_border: "#HEX"  # Response box border color
+  #   spinner:
+  #     waiting_faces: ["(⚔)", "(⛨)"]       # Faces shown while waiting
+  #     thinking_faces: ["(⚔)", "(⌁)"]      # Faces shown while thinking
+  #     thinking_verbs: ["forging", "plotting"]  # Verbs for spinner messages
+  #     wings:                                # Optional left/right spinner decorations
+  #       - ["⟪⚔", "⚔⟫"]
+  #       - ["⟪▲", "▲⟫"]
+  #   branding:
+  #     agent_name: "My Agent"               # Banner title and branding
+  #     welcome: "Welcome message"           # Shown at CLI startup
+  #     response_label: " ⚔ Agent "         # Response box header label
+  #     prompt_symbol: "⚔ ❯ "              # Prompt symbol
+  #   tool_prefix: "╎"                       # Tool output line prefix (default: ┊)
+  #
+  skin: default
--- a/cli.py
+++ b/cli.py
@@ -19,6 +19,7 @@ import sys
 import json
 import atexit
 import uuid
+import textwrap
 from pathlib import Path
 from datetime import datetime
 from typing import List, Dict, Any, Optional
@@ -45,6 +46,11 @@ from prompt_toolkit.widgets import TextArea
 from prompt_toolkit.key_binding import KeyBindings
 from prompt_toolkit import print_formatted_text as _pt_print
 from prompt_toolkit.formatted_text import ANSI as _PT_ANSI
+try:
+    from prompt_toolkit.cursor_shapes import CursorShape
+    _STEADY_CURSOR = CursorShape.BLOCK  # Non-blinking block cursor
+except (ImportError, AttributeError):
+    _STEADY_CURSOR = None
 import threading
 import queue

@@ -158,9 +164,11 @@ def load_cli_config() -> Dict[str, Any]:
            "singularity_image": "docker://python:3.11",
            "modal_image": "python:3.11",
            "daytona_image": "nikolaik/python-nodejs:python3.11-nodejs20",
+            "docker_volumes": [],  # host:container volume mounts for Docker backend
        },
        "browser": {
            "inactivity_timeout": 120,  # Auto-cleanup inactive browser sessions after 2 min
+            "record_sessions": False,  # Auto-record browser sessions as WebM videos
        },
        "compression": {
            "enabled": True,      # Auto-compress when approaching context limit
@@ -194,6 +202,7 @@ def load_cli_config() -> Dict[str, Any]:
        "display": {
            "compact": False,
            "resume_display": "full",
+            "skin": "default",
        },
        "clarify": {
            "timeout": 120,  # Seconds to wait for a clarify answer before auto-proceeding
@@ -248,8 +257,13 @@ def load_cli_config() -> Dict[str, Any]:
                if key not in defaults and key != "model":
                    defaults[key] = file_config[key]
            
-            # Handle root-level max_turns (backwards compat) - copy to agent.max_turns
-            if "max_turns" in file_config and "agent" not in file_config:
+            # Handle legacy root-level max_turns (backwards compat) - copy to
+            # agent.max_turns whenever the nested key is missing.
+            agent_file_config = file_config.get("agent")
+            if "max_turns" in file_config and not (
+                isinstance(agent_file_config, dict)
+                and agent_file_config.get("max_turns") is not None
+            ):
                defaults["agent"]["max_turns"] = file_config["max_turns"]
        except Exception as e:
            logger.warning("Failed to load cli-config.yaml: %s", e)
@@ -363,11 +377,26 @@ def load_cli_config() -> Dict[str, Any]:
        if model:
            os.environ[model_env] = model
    
+    # Security settings
+    security_config = defaults.get("security", {})
+    if isinstance(security_config, dict):
+        redact = security_config.get("redact_secrets")
+        if redact is not None:
+            os.environ["HERMES_REDACT_SECRETS"] = str(redact).lower()
+
    return defaults

 # Load configuration at module startup
 CLI_CONFIG = load_cli_config()

+# Initialize the skin engine from config
+try:
+    from hermes_cli.skin_engine import init_skin_from_config
+    init_skin_from_config(CLI_CONFIG)
+except Exception:
+    pass  # Skin engine is optional — default skin used if unavailable
+
+from rich import box as rich_box
 from rich.console import Console
 from rich.panel import Panel
 from rich.table import Table
@@ -686,6 +715,8 @@ class ChatConsole:
    def print(self, *args, **kwargs):
        self._buffer.seek(0)
        self._buffer.truncate()
+        # Read terminal width at render time so panels adapt to current size
+        self._inner.width = shutil.get_terminal_size((80, 24)).columns
        self._inner.print(*args, **kwargs)
        output = self._buffer.getvalue()
        for line in output.rstrip("\n").split("\n"):
@@ -717,6 +748,7 @@ HERMES_CADUCEUS = """[#CD7F32]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡀⠀⣀⣀
 [#B8860B]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]"""

 # Compact banner for smaller terminals (fallback)
+# Note: built dynamically by _build_compact_banner() to fit terminal width
 COMPACT_BANNER = """
 [bold #FFD700]╔══════════════════════════════════════════════════════════════╗[/]
 [bold #FFD700]║[/]  [#FFBF00]⚕ NOUS HERMES[/] [dim #B8860B]- AI Agent Framework[/]              [bold #FFD700]║[/]
@@ -725,6 +757,26 @@ COMPACT_BANNER = """
 """


+def _build_compact_banner() -> str:
+    """Build a compact banner that fits the current terminal width."""
+    w = min(shutil.get_terminal_size().columns - 2, 64)
+    if w < 30:
+        return "\n[#FFBF00]⚕ NOUS HERMES[/] [dim #B8860B]- Nous Research[/]\n"
+    inner = w - 2  # inside the box border
+    bar = "═" * w
+    line1 = "⚕ NOUS HERMES - AI Agent Framework"
+    line2 = "Messenger of the Digital Gods  ·  Nous Research"
+    # Truncate and pad to fit
+    line1 = line1[:inner - 2].ljust(inner - 2)
+    line2 = line2[:inner - 2].ljust(inner - 2)
+    return (
+        f"\n[bold #FFD700]╔{bar}╗[/]\n"
+        f"[bold #FFD700]║[/] [#FFBF00]{line1}[/] [bold #FFD700]║[/]\n"
+        f"[bold #FFD700]║[/] [dim #B8860B]{line2}[/] [bold #FFD700]║[/]\n"
+        f"[bold #FFD700]╚{bar}╝[/]\n"
+    )
+
+
 def _get_available_skills() -> Dict[str, List[str]]:
    """
    Scan ~/.hermes/skills/ and return skills grouped by category.
@@ -798,25 +850,43 @@ def build_welcome_banner(console: Console, model: str, cwd: str, tools: List[dic
    layout_table.add_column("right", justify="left")
    
    # Build left content: caduceus + model info
-    left_lines = ["", HERMES_CADUCEUS, ""]
+    # Resolve skin colors for the banner
+    try:
+        from hermes_cli.skin_engine import get_active_skin
+        _bskin = get_active_skin()
+        _accent = _bskin.get_color("banner_accent", "#FFBF00")
+        _dim = _bskin.get_color("banner_dim", "#B8860B")
+        _text = _bskin.get_color("banner_text", "#FFF8DC")
+        _session_c = _bskin.get_color("session_border", "#8B8682")
+        _title_c = _bskin.get_color("banner_title", "#FFD700")
+        _border_c = _bskin.get_color("banner_border", "#CD7F32")
+        _agent_name = _bskin.get_branding("agent_name", "Hermes Agent")
+    except Exception:
+        _bskin = None
+        _accent, _dim, _text = "#FFBF00", "#B8860B", "#FFF8DC"
+        _session_c, _title_c, _border_c = "#8B8682", "#FFD700", "#CD7F32"
+        _agent_name = "Hermes Agent"
+
+    _hero = _bskin.banner_hero if hasattr(_bskin, 'banner_hero') and _bskin.banner_hero else HERMES_CADUCEUS
+    left_lines = ["", _hero, ""]
    
    # Shorten model name for display
    model_short = model.split("/")[-1] if "/" in model else model
    if len(model_short) > 28:
        model_short = model_short[:25] + "..."
    
-    ctx_str = f" [dim #B8860B]·[/] [dim #B8860B]{_format_context_length(context_length)} context[/]" if context_length else ""
-    left_lines.append(f"[#FFBF00]{model_short}[/]{ctx_str} [dim #B8860B]·[/] [dim #B8860B]Nous Research[/]")
-    left_lines.append(f"[dim #B8860B]{cwd}[/]")
+    ctx_str = f" [dim {_dim}]·[/] [dim {_dim}]{_format_context_length(context_length)} context[/]" if context_length else ""
+    left_lines.append(f"[{_accent}]{model_short}[/]{ctx_str} [dim {_dim}]·[/] [dim {_dim}]Nous Research[/]")
+    left_lines.append(f"[dim {_dim}]{cwd}[/]")
    
    # Add session ID if provided
    if session_id:
-        left_lines.append(f"[dim #8B8682]Session: {session_id}[/]")
+        left_lines.append(f"[dim {_session_c}]Session: {session_id}[/]")
    left_content = "\n".join(left_lines)
    
    # Build right content: tools list grouped by toolset
    right_lines = []
-    right_lines.append("[bold #FFBF00]Available Tools[/]")
+    right_lines.append(f"[bold {_accent}]Available Tools[/]")
    
    # Group tools by toolset (include all possible tools, both enabled and disabled)
    toolsets_dict = {}
@@ -853,7 +923,7 @@ def build_welcome_banner(console: Console, model: str, cwd: str, tools: List[dic
            if name in disabled_tools:
                colored_names.append(f"[red]{name}[/]")
            else:
-                colored_names.append(f"[#FFF8DC]{name}[/]")
+                colored_names.append(f"[{_text}]{name}[/]")
        
        tools_str = ", ".join(colored_names)
        # Truncate if too long (accounting for markup)
@@ -875,18 +945,18 @@ def build_welcome_banner(console: Console, model: str, cwd: str, tools: List[dic
                elif name in disabled_tools:
                    colored_names.append(f"[red]{name}[/]")
                else:
-                    colored_names.append(f"[#FFF8DC]{name}[/]")
+                    colored_names.append(f"[{_text}]{name}[/]")
            tools_str = ", ".join(colored_names)
        
-        right_lines.append(f"[dim #B8860B]{toolset}:[/] {tools_str}")
+        right_lines.append(f"[dim {_dim}]{toolset}:[/] {tools_str}")
    
    if remaining_toolsets > 0:
-        right_lines.append(f"[dim #B8860B](and {remaining_toolsets} more toolsets...)[/]")
+        right_lines.append(f"[dim {_dim}](and {remaining_toolsets} more toolsets...)[/]")
    
    right_lines.append("")
    
    # Add skills section
-    right_lines.append("[bold #FFBF00]Available Skills[/]")
+    right_lines.append(f"[bold {_accent}]Available Skills[/]")
    skills_by_category = _get_available_skills()
    total_skills = sum(len(s) for s in skills_by_category.values())
    
@@ -902,12 +972,12 @@ def build_welcome_banner(console: Console, model: str, cwd: str, tools: List[dic
            # Truncate if still too long
            if len(skills_str) > 50:
                skills_str = skills_str[:47] + "..."
-            right_lines.append(f"[dim #B8860B]{category}:[/] [#FFF8DC]{skills_str}[/]")
+            right_lines.append(f"[dim {_dim}]{category}:[/] [{_text}]{skills_str}[/]")
    else:
-        right_lines.append("[dim #B8860B]No skills installed[/]")
+        right_lines.append(f"[dim {_dim}]No skills installed[/]")
    
    right_lines.append("")
-    right_lines.append(f"[dim #B8860B]{len(tools)} tools · {total_skills} skills · /help for commands[/]")
+    right_lines.append(f"[dim {_dim}]{len(tools)} tools · {total_skills} skills · /help for commands[/]")
    
    right_content = "\n".join(right_lines)
    
@@ -917,15 +987,18 @@ def build_welcome_banner(console: Console, model: str, cwd: str, tools: List[dic
    # Wrap in a panel with the title
    outer_panel = Panel(
        layout_table,
-        title=f"[bold #FFD700]Hermes Agent {VERSION}[/]",
-        border_style="#CD7F32",
+        title=f"[bold {_title_c}]{_agent_name} {VERSION}[/]",
+        border_style=_border_c,
        padding=(0, 2),
    )
    
-    # Print the big HERMES-AGENT logo first (no panel wrapper for full width)
-    console.print()
-    console.print(HERMES_AGENT_LOGO)
+    # Print the big logo — use skin's custom logo if available
    console.print()
+    term_width = shutil.get_terminal_size().columns
+    if term_width >= 95:
+        _logo = _bskin.banner_logo if hasattr(_bskin, 'banner_logo') and _bskin.banner_logo else HERMES_AGENT_LOGO
+        console.print(_logo)
+        console.print()
    
    # Print the panel with caduceus and info
    console.print(outer_panel)
@@ -1013,6 +1086,7 @@ class HermesCLI:
        verbose: bool = False,
        compact: bool = False,
        resume: str = None,
+        checkpoints: bool = False,
    ):
        """
        Initialize the Hermes CLI.
@@ -1094,6 +1168,13 @@ class HermesCLI:
            if invalid:
                self.console.print(f"[bold red]Warning: Unknown toolsets: {', '.join(invalid)}[/]")
        
+        # Filesystem checkpoints: CLI flag > config
+        cp_cfg = CLI_CONFIG.get("checkpoints", {})
+        if isinstance(cp_cfg, bool):
+            cp_cfg = {"enabled": cp_cfg}
+        self.checkpoints_enabled = checkpoints or cp_cfg.get("enabled", False)
+        self.checkpoint_max_snapshots = cp_cfg.get("max_snapshots", 50)
+        
        # Ephemeral system prompt: env var takes precedence, then config
        self.system_prompt = (
            os.getenv("HERMES_EPHEMERAL_SYSTEM_PROMPT", "")
@@ -1120,6 +1201,10 @@ class HermesCLI:
        self._provider_require_params = pr.get("require_parameters", False)
        self._provider_data_collection = pr.get("data_collection")
        
+        # Fallback model config — tried when primary provider fails after retries
+        fb = CLI_CONFIG.get("fallback_model") or {}
+        self._fallback_model = fb if fb.get("provider") and fb.get("model") else None
+
        # Agent will be initialized on first use
        self.agent: Optional[AIAgent] = None
        self._app = None  # prompt_toolkit Application (set in run())
@@ -1151,6 +1236,7 @@ class HermesCLI:
        # History file for persistent input recall across sessions
        self._history_file = Path.home() / ".hermes_history"
        self._last_invalidate: float = 0.0  # throttle UI repaints
+        self._spinner_text: str = ""  # thinking spinner text for TUI

    def _invalidate(self, min_interval: float = 0.25) -> None:
        """Throttled UI repaint — prevents terminal blinking on slow/SSH connections."""
@@ -1214,6 +1300,11 @@ class HermesCLI:

        return changed

+    def _on_thinking(self, text: str) -> None:
+        """Called by agent when thinking starts/stops. Updates TUI spinner."""
+        self._spinner_text = text or ""
+        self._invalidate()
+
    def _ensure_runtime_credentials(self) -> bool:
        """
        Ensure runtime credentials are resolved before agent use.
@@ -1351,6 +1442,10 @@ class HermesCLI:
                session_db=self._session_db,
                clarify_callback=self._clarify_callback,
                honcho_session_key=self.session_id,
+                fallback_model=self._fallback_model,
+                thinking_callback=self._on_thinking,
+                checkpoints_enabled=self.checkpoints_enabled,
+                checkpoint_max_snapshots=self.checkpoint_max_snapshots,
            )
            # Apply any pending title now that the session exists in the DB
            if self._pending_title and self._session_db:
@@ -1370,8 +1465,13 @@ class HermesCLI:
        """Display the welcome banner in Claude Code style."""
        self.console.clear()
        
-        if self.compact:
-            self.console.print(COMPACT_BANNER)
+        # Auto-compact for narrow terminals — the full banner with caduceus
+        # + tool list needs ~80 columns minimum to render without wrapping.
+        term_width = shutil.get_terminal_size().columns
+        use_compact = self.compact or term_width < 80
+        
+        if use_compact:
+            self.console.print(_build_compact_banner())
            self._show_status()
        else:
            # Get tools for display
@@ -1615,6 +1715,55 @@ class HermesCLI:
        self._image_counter -= 1
        return False

+    def _handle_rollback_command(self, command: str):
+        """Handle /rollback — list or restore filesystem checkpoints."""
+        from tools.checkpoint_manager import CheckpointManager, format_checkpoint_list
+
+        if not hasattr(self, 'agent') or not self.agent:
+            print("  No active agent session.")
+            return
+
+        mgr = self.agent._checkpoint_mgr
+        if not mgr.enabled:
+            print("  Checkpoints are not enabled.")
+            print("  Enable with: hermes --checkpoints")
+            print("  Or in config.yaml: checkpoints: { enabled: true }")
+            return
+
+        cwd = os.getenv("TERMINAL_CWD", os.getcwd())
+        parts = command.split(maxsplit=1)
+        arg = parts[1].strip() if len(parts) > 1 else ""
+
+        if not arg:
+            # List checkpoints
+            checkpoints = mgr.list_checkpoints(cwd)
+            print(format_checkpoint_list(checkpoints, cwd))
+        else:
+            # Restore by number or hash
+            checkpoints = mgr.list_checkpoints(cwd)
+            if not checkpoints:
+                print(f"  No checkpoints found for {cwd}")
+                return
+
+            target_hash = None
+            try:
+                idx = int(arg) - 1  # 1-indexed for user
+                if 0 <= idx < len(checkpoints):
+                    target_hash = checkpoints[idx]["hash"]
+                else:
+                    print(f"  Invalid checkpoint number. Use 1-{len(checkpoints)}.")
+                    return
+            except ValueError:
+                # Try as a git hash
+                target_hash = arg
+
+            result = mgr.restore(cwd, target_hash)
+            if result["success"]:
+                print(f"  ✅ Restored to checkpoint {result['restored_to']}: {result['reason']}")
+                print(f"  A pre-rollback snapshot was saved automatically.")
+            else:
+                print(f"  ❌ {result['error']}")
+
    def _handle_paste_command(self):
        """Handle /paste — explicitly check clipboard for an image.

@@ -2381,8 +2530,9 @@ class HermesCLI:
            # and gets mangled by patch_stdout).
            if self._app:
                cc = ChatConsole()
-                if self.compact:
-                    cc.print(COMPACT_BANNER)
+                term_w = shutil.get_terminal_size().columns
+                if self.compact or term_w < 80:
+                    cc.print(_build_compact_banner())
                else:
                    tools = get_tool_definitions(enabled_toolsets=self.enabled_toolsets, quiet_mode=True)
                    cwd = os.getenv("TERMINAL_CWD", os.getcwd())
@@ -2623,6 +2773,10 @@ class HermesCLI:
            self._handle_paste_command()
        elif cmd_lower == "/reload-mcp":
            self._reload_mcp()
+        elif cmd_lower.startswith("/rollback"):
+            self._handle_rollback_command(cmd_original)
+        elif cmd_lower.startswith("/skin"):
+            self._handle_skin_command(cmd_original)
        else:
            # Check for skill slash commands (/gif-search, /axolotl, etc.)
            base_cmd = cmd_lower.split()[0]
@@ -2642,6 +2796,43 @@ class HermesCLI:
        
        return True
    
+    def _handle_skin_command(self, cmd: str):
+        """Handle /skin [name] — show or change the display skin."""
+        try:
+            from hermes_cli.skin_engine import list_skins, set_active_skin, get_active_skin_name
+        except ImportError:
+            print("Skin engine not available.")
+            return
+
+        parts = cmd.strip().split(maxsplit=1)
+        if len(parts) < 2 or not parts[1].strip():
+            # Show current skin and list available
+            current = get_active_skin_name()
+            skins = list_skins()
+            print(f"\n  Current skin: {current}")
+            print(f"  Available skins:")
+            for s in skins:
+                marker = " ●" if s["name"] == current else "  "
+                source = f" ({s['source']})" if s["source"] == "user" else ""
+                print(f"   {marker} {s['name']}{source} — {s['description']}")
+            print(f"\n  Usage: /skin <name>")
+            print(f"  Custom skins: drop a YAML file in ~/.hermes/skins/\n")
+            return
+
+        new_skin = parts[1].strip().lower()
+        available = {s["name"] for s in list_skins()}
+        if new_skin not in available:
+            print(f"  Unknown skin: {new_skin}")
+            print(f"  Available: {', '.join(sorted(available))}")
+            return
+
+        set_active_skin(new_skin)
+        if save_config_value("display.skin", new_skin):
+            print(f"  Skin set to: {new_skin} (saved)")
+        else:
+            print(f"  Skin set to: {new_skin}")
+        print("  Note: banner colors will update on next session start.")
+
    def _toggle_verbose(self):
        """Cycle tool progress mode: off → new → all → verbose → off."""
        cycle = ["off", "new", "all", "verbose"]
@@ -2890,8 +3081,16 @@ class HermesCLI:
        # Trigger prompt_toolkit repaint from this (non-main) thread
        self._invalidate()

-        # Poll in 1-second ticks so the countdown refreshes in the UI.
-        # Each tick triggers an invalidate() to repaint the hint line.
+        # Poll for the user's response.  The countdown in the hint line
+        # updates on each invalidate — but frequent repaints cause visible
+        # flicker in some terminals (Kitty, ghostty).  We only refresh the
+        # countdown every 5 s; selection changes (↑/↓) trigger instant
+        # Poll for the user's response.  The countdown in the hint line
+        # updates on each invalidate — but frequent repaints cause visible
+        # flicker in some terminals (Kitty, ghostty).  We only refresh the
+        # countdown every 5 s; selection changes (↑/↓) trigger instant
+        # repaints via the key bindings.
+        _last_countdown_refresh = _time.monotonic()
        while True:
            try:
                result = response_queue.get(timeout=1)
@@ -2901,8 +3100,14 @@ class HermesCLI:
                remaining = self._clarify_deadline - _time.monotonic()
                if remaining <= 0:
                    break
-                # Repaint so the countdown updates
-                self._invalidate()
+                # Only repaint every 5 s for the countdown — avoids flicker
+                now = _time.monotonic()
+                if now - _last_countdown_refresh >= 5.0:
+                    _last_countdown_refresh = now
+                    self._invalidate()
+                if now - _last_countdown_refresh >= 5.0:
+                    _last_countdown_refresh = now
+                    self._invalidate()

        # Timed out — tear down the UI and let the agent decide
        self._clarify_state = None
@@ -2982,6 +3187,9 @@ class HermesCLI:

        self._invalidate()

+        # Same throttled countdown as _clarify_callback — repaint only
+        # every 5 s to avoid flicker in Kitty / ghostty / etc.
+        _last_countdown_refresh = _time.monotonic()
        while True:
            try:
                result = response_queue.get(timeout=1)
@@ -2993,11 +3201,16 @@ class HermesCLI:
                remaining = self._approval_deadline - _time.monotonic()
                if remaining <= 0:
                    break
-                self._invalidate()
+                now = _time.monotonic()
+                if now - _last_countdown_refresh >= 5.0:
+                    _last_countdown_refresh = now
+                    self._invalidate()

        self._approval_state = None
        self._approval_deadline = 0
        self._invalidate()
+        return "deny"
+
    def chat(self, message, images: list = None) -> Optional[str]:
        """
        Send a message to the agent and get a response.
@@ -3036,8 +3249,7 @@ class HermesCLI:
        # Add user message to history
        self.conversation_history.append({"role": "user", "content": message})
        
-        w = shutil.get_terminal_size().columns
-        _cprint(f"{_GOLD}{'─' * w}{_RST}")
+        _cprint(f"{_GOLD}{'─' * 40}{_RST}")
        print(flush=True)
        
        try:
@@ -3112,15 +3324,26 @@ class HermesCLI:
                    response = response + "\n\n---\n_[Interrupted - processing new message]_"
            
            if response:
-                w = shutil.get_terminal_size().columns
-                label = " ⚕ Hermes "
-                fill = w - 2 - len(label)  # 2 for ╭ and ╮
-                top = f"{_GOLD}╭─{label}{'─' * max(fill - 1, 0)}╮{_RST}"
-                bot = f"{_GOLD}╰{'─' * (w - 2)}╯{_RST}"
+                # Use a Rich Panel for the response box — adapts to terminal
+                # width at render time instead of hard-coding border length.
+                try:
+                    from hermes_cli.skin_engine import get_active_skin
+                    _skin = get_active_skin()
+                    label = _skin.get_branding("response_label", "⚕ Hermes")
+                    _resp_color = _skin.get_color("response_border", "#CD7F32")
+                except Exception:
+                    label = "⚕ Hermes"
+                    _resp_color = "#CD7F32"

-                # Render box + response as a single _cprint call so
-                # nothing can interleave between the box borders.
-                _cprint(f"\n{top}\n{response}\n\n{bot}")
+                _chat_console = ChatConsole()
+                _chat_console.print(Panel(
+                    response,
+                    title=f"[bold]{label}[/bold]",
+                    title_align="left",
+                    border_style=_resp_color,
+                    box=rich_box.HORIZONTALS,
+                    padding=(1, 2),
+                ))
            
            # Play terminal bell when agent finishes (if enabled).
            # Works over SSH — the bell propagates to the user's terminal.
@@ -3185,7 +3408,15 @@ class HermesCLI:
            if self._preload_resumed_session():
                self._display_resumed_history()

-        self.console.print("[#FFF8DC]Welcome to Hermes Agent! Type your message or /help for commands.[/]")
+        try:
+            from hermes_cli.skin_engine import get_active_skin
+            _welcome_skin = get_active_skin()
+            _welcome_text = _welcome_skin.get_branding("welcome", "Welcome to Hermes Agent! Type your message or /help for commands.")
+            _welcome_color = _welcome_skin.get_color("banner_text", "#FFF8DC")
+        except Exception:
+            _welcome_text = "Welcome to Hermes Agent! Type your message or /help for commands."
+            _welcome_color = "#FFF8DC"
+        self.console.print(f"[{_welcome_color}]{_welcome_text}[/]")
        self.console.print()
        
        # State for async operation
@@ -3573,6 +3804,8 @@ class HermesCLI:
                return "type password (hidden), Enter to skip"
            if cli_ref._approval_state:
                return ""
+            if cli_ref._clarify_freetext:
+                return "type your answer here and press Enter"
            if cli_ref._clarify_state:
                return ""
            if cli_ref._agent_running:
@@ -3623,6 +3856,20 @@ class HermesCLI:
            # right up against the top rule of the input area
            return 1 if cli_ref._agent_running else 0

+        def get_spinner_text():
+            txt = cli_ref._spinner_text
+            if not txt:
+                return []
+            return [('class:hint', f'  {txt}')]
+
+        def get_spinner_height():
+            return 1 if cli_ref._spinner_text else 0
+
+        spinner_widget = Window(
+            content=FormattedTextControl(get_spinner_text),
+            height=get_spinner_height,
+        )
+
        spacer = Window(
            content=FormattedTextControl(get_hint_text),
            height=get_hint_height,
@@ -3630,6 +3877,32 @@ class HermesCLI:

        # --- Clarify tool: dynamic display widget for questions + choices ---

+        def _panel_box_width(title: str, content_lines: list[str], min_width: int = 46, max_width: int = 76) -> int:
+            """Choose a stable panel width wide enough for the title and content."""
+            term_cols = shutil.get_terminal_size((100, 20)).columns
+            longest = max([len(title)] + [len(line) for line in content_lines] + [min_width - 4])
+            inner = min(max(longest + 4, min_width - 2), max_width - 2, max(24, term_cols - 6))
+            return inner + 2  # account for the single leading/trailing spaces inside borders
+
+        def _wrap_panel_text(text: str, width: int, subsequent_indent: str = "") -> list[str]:
+            wrapped = textwrap.wrap(
+                text,
+                width=max(8, width),
+                break_long_words=False,
+                break_on_hyphens=False,
+                subsequent_indent=subsequent_indent,
+            )
+            return wrapped or [""]
+
+        def _append_panel_line(lines, border_style: str, content_style: str, text: str, box_width: int) -> None:
+            inner_width = max(0, box_width - 2)
+            lines.append((border_style, "│ "))
+            lines.append((content_style, text.ljust(inner_width)))
+            lines.append((border_style, " │\n"))
+
+        def _append_blank_panel_line(lines, border_style: str, box_width: int) -> None:
+            lines.append((border_style, "│" + (" " * box_width) + "│\n"))
+
        def _get_clarify_display():
            """Build styled text for the clarify question/choices panel."""
            state = cli_ref._clarify_state
@@ -3639,43 +3912,62 @@ class HermesCLI:
            question = state["question"]
            choices = state.get("choices") or []
            selected = state.get("selected", 0)
+            preview_lines = _wrap_panel_text(question, 60)
+            for i, choice in enumerate(choices):
+                prefix = "❯ " if i == selected and not cli_ref._clarify_freetext else "  "
+                preview_lines.extend(_wrap_panel_text(f"{prefix}{choice}", 60, subsequent_indent="  "))
+            other_label = (
+                "❯ Other (type below)" if cli_ref._clarify_freetext
+                else "❯ Other (type your answer)" if selected == len(choices)
+                else "  Other (type your answer)"
+            )
+            preview_lines.extend(_wrap_panel_text(other_label, 60, subsequent_indent="  "))
+            box_width = _panel_box_width("Hermes needs your input", preview_lines)
+            inner_text_width = max(8, box_width - 2)

            lines = []
            # Box top border
            lines.append(('class:clarify-border', '╭─ '))
            lines.append(('class:clarify-title', 'Hermes needs your input'))
-            lines.append(('class:clarify-border', ' ─────────────────────────────╮\n'))
-            lines.append(('class:clarify-border', '│\n'))
+            lines.append(('class:clarify-border', ' ' + ('─' * max(0, box_width - len("Hermes needs your input") - 3)) + '╮\n'))
+            _append_blank_panel_line(lines, 'class:clarify-border', box_width)

            # Question text
-            lines.append(('class:clarify-border', '│  '))
-            lines.append(('class:clarify-question', question))
-            lines.append(('', '\n'))
-            lines.append(('class:clarify-border', '│\n'))
+            for wrapped in _wrap_panel_text(question, inner_text_width):
+                _append_panel_line(lines, 'class:clarify-border', 'class:clarify-question', wrapped, box_width)
+            _append_blank_panel_line(lines, 'class:clarify-border', box_width)
+
+            if cli_ref._clarify_freetext and not choices:
+                guidance = "Type your answer in the prompt below, then press Enter."
+                for wrapped in _wrap_panel_text(guidance, inner_text_width):
+                    _append_panel_line(lines, 'class:clarify-border', 'class:clarify-choice', wrapped, box_width)
+                _append_blank_panel_line(lines, 'class:clarify-border', box_width)

            if choices:
                # Multiple-choice mode: show selectable options
                for i, choice in enumerate(choices):
-                    lines.append(('class:clarify-border', '│  '))
-                    if i == selected and not cli_ref._clarify_freetext:
-                        lines.append(('class:clarify-selected', f'❯ {choice}'))
-                    else:
-                        lines.append(('class:clarify-choice', f'  {choice}'))
-                    lines.append(('', '\n'))
+                    style = 'class:clarify-selected' if i == selected and not cli_ref._clarify_freetext else 'class:clarify-choice'
+                    prefix = '❯ ' if i == selected and not cli_ref._clarify_freetext else '  '
+                    wrapped_lines = _wrap_panel_text(f"{prefix}{choice}", inner_text_width, subsequent_indent="  ")
+                    for wrapped in wrapped_lines:
+                        _append_panel_line(lines, 'class:clarify-border', style, wrapped, box_width)

                # "Other" option (5th line, only shown when choices exist)
                other_idx = len(choices)
-                lines.append(('class:clarify-border', '│  '))
                if selected == other_idx and not cli_ref._clarify_freetext:
-                    lines.append(('class:clarify-selected', '❯ Other (type your answer)'))
+                    other_style = 'class:clarify-selected'
+                    other_label = '❯ Other (type your answer)'
                elif cli_ref._clarify_freetext:
-                    lines.append(('class:clarify-active-other', '❯ Other (type below)'))
+                    other_style = 'class:clarify-active-other'
+                    other_label = '❯ Other (type below)'
                else:
-                    lines.append(('class:clarify-choice', '  Other (type your answer)'))
-                lines.append(('', '\n'))
+                    other_style = 'class:clarify-choice'
+                    other_label = '  Other (type your answer)'
+                for wrapped in _wrap_panel_text(other_label, inner_text_width, subsequent_indent="  "):
+                    _append_panel_line(lines, 'class:clarify-border', other_style, wrapped, box_width)

-            lines.append(('class:clarify-border', '│\n'))
-            lines.append(('class:clarify-border', '╰──────────────────────────────────────────────────╯\n'))
+            _append_blank_panel_line(lines, 'class:clarify-border', box_width)
+            lines.append(('class:clarify-border', '╰' + ('─' * box_width) + '╯\n'))
            return lines

        clarify_widget = ConditionalContainer(
@@ -3692,16 +3984,18 @@ class HermesCLI:
            state = cli_ref._sudo_state
            if not state:
                return []
+            title = '🔐 Sudo Password Required'
+            body = 'Enter password below (hidden), or press Enter to skip'
+            box_width = _panel_box_width(title, [body])
+            inner = max(0, box_width - 2)
            lines = []
            lines.append(('class:sudo-border', '╭─ '))
-            lines.append(('class:sudo-title', '🔐 Sudo Password Required'))
-            lines.append(('class:sudo-border', ' ──────────────────────────╮\n'))
-            lines.append(('class:sudo-border', '│\n'))
-            lines.append(('class:sudo-border', '│  '))
-            lines.append(('class:sudo-text', 'Enter password below (hidden), or press Enter to skip'))
-            lines.append(('', '\n'))
-            lines.append(('class:sudo-border', '│\n'))
-            lines.append(('class:sudo-border', '╰──────────────────────────────────────────────────╯\n'))
+            lines.append(('class:sudo-title', title))
+            lines.append(('class:sudo-border', ' ' + ('─' * max(0, box_width - len(title) - 3)) + '╮\n'))
+            _append_blank_panel_line(lines, 'class:sudo-border', box_width)
+            _append_panel_line(lines, 'class:sudo-border', 'class:sudo-text', body, box_width)
+            _append_blank_panel_line(lines, 'class:sudo-border', box_width)
+            lines.append(('class:sudo-border', '╰' + ('─' * box_width) + '╯\n'))
            return lines

        sudo_widget = ConditionalContainer(
@@ -3730,29 +4024,32 @@ class HermesCLI:
                "always": "Add to permanent allowlist",
                "deny": "Deny",
            }
+            preview_lines = _wrap_panel_text(description, 60)
+            preview_lines.extend(_wrap_panel_text(cmd_display, 60))
+            for i, choice in enumerate(choices):
+                prefix = '❯ ' if i == selected else '  '
+                preview_lines.extend(_wrap_panel_text(f"{prefix}{choice_labels.get(choice, choice)}", 60, subsequent_indent="  "))
+            box_width = _panel_box_width("⚠️  Dangerous Command", preview_lines)
+            inner_text_width = max(8, box_width - 2)

            lines = []
            lines.append(('class:approval-border', '╭─ '))
            lines.append(('class:approval-title', '⚠️  Dangerous Command'))
-            lines.append(('class:approval-border', ' ───────────────────────────────╮\n'))
-            lines.append(('class:approval-border', '│\n'))
-            lines.append(('class:approval-border', '│  '))
-            lines.append(('class:approval-desc', description))
-            lines.append(('', '\n'))
-            lines.append(('class:approval-border', '│  '))
-            lines.append(('class:approval-cmd', cmd_display))
-            lines.append(('', '\n'))
-            lines.append(('class:approval-border', '│\n'))
+            lines.append(('class:approval-border', ' ' + ('─' * max(0, box_width - len("⚠️  Dangerous Command") - 3)) + '╮\n'))
+            _append_blank_panel_line(lines, 'class:approval-border', box_width)
+            for wrapped in _wrap_panel_text(description, inner_text_width):
+                _append_panel_line(lines, 'class:approval-border', 'class:approval-desc', wrapped, box_width)
+            for wrapped in _wrap_panel_text(cmd_display, inner_text_width):
+                _append_panel_line(lines, 'class:approval-border', 'class:approval-cmd', wrapped, box_width)
+            _append_blank_panel_line(lines, 'class:approval-border', box_width)
            for i, choice in enumerate(choices):
-                lines.append(('class:approval-border', '│  '))
                label = choice_labels.get(choice, choice)
-                if i == selected:
-                    lines.append(('class:approval-selected', f'❯ {label}'))
-                else:
-                    lines.append(('class:approval-choice', f'  {label}'))
-                lines.append(('', '\n'))
-            lines.append(('class:approval-border', '│\n'))
-            lines.append(('class:approval-border', '╰──────────────────────────────────────────────────────╯\n'))
+                style = 'class:approval-selected' if i == selected else 'class:approval-choice'
+                prefix = '❯ ' if i == selected else '  '
+                for wrapped in _wrap_panel_text(f"{prefix}{label}", inner_text_width, subsequent_indent="  "):
+                    _append_panel_line(lines, 'class:approval-border', style, wrapped, box_width)
+            _append_blank_panel_line(lines, 'class:approval-border', box_width)
+            lines.append(('class:approval-border', '╰' + ('─' * box_width) + '╯\n'))
            return lines

        approval_widget = ConditionalContainer(
@@ -3805,6 +4102,7 @@ class HermesCLI:
                sudo_widget,
                approval_widget,
                clarify_widget,
+                spinner_widget,
                spacer,
                input_rule_top,
                image_bar,
@@ -3859,6 +4157,7 @@ class HermesCLI:
            style=style,
            full_screen=False,
            mouse_support=False,
+            **({'cursor': _STEADY_CURSOR} if _STEADY_CURSOR is not None else {}),
        )
        self._app = app  # Store reference for clarify_callback
        
@@ -3927,6 +4226,7 @@ class HermesCLI:
                        self.chat(user_input, images=submit_images or None)
                    finally:
                        self._agent_running = False
+                        self._spinner_text = ""
                        app.invalidate()  # Refresh status line
                    
                except Exception as e:
@@ -3987,6 +4287,7 @@ def main(
    resume: str = None,
    worktree: bool = False,
    w: bool = False,
+    checkpoints: bool = False,
 ):
    """
    Hermes Agent CLI - Interactive AI Assistant
@@ -4091,6 +4392,7 @@ def main(
        verbose=verbose,
        compact=compact,
        resume=resume,
+        checkpoints=checkpoints,
    )

    # Inject worktree context into agent's system prompt
--- a/cron/jobs.py
+++ b/cron/jobs.py
@@ -26,7 +26,7 @@ except ImportError:
 # Configuration
 # =============================================================================

-HERMES_DIR = Path.home() / ".hermes"
+HERMES_DIR = Path(os.getenv("HERMES_HOME", Path.home() / ".hermes"))
 CRON_DIR = HERMES_DIR / "cron"
 JOBS_FILE = CRON_DIR / "jobs.json"
 OUTPUT_DIR = CRON_DIR / "output"
--- a/cron/scheduler.py
+++ b/cron/scheduler.py
@@ -98,6 +98,7 @@ def _deliver_result(job: dict, content: str) -> None:
        "discord": Platform.DISCORD,
        "slack": Platform.SLACK,
        "whatsapp": Platform.WHATSAPP,
+        "signal": Platform.SIGNAL,
    }
    platform = platform_map.get(platform_name.lower())
    if not platform:
--- a/datagen-config-examples/web_research.yaml
+++ b/datagen-config-examples/web_research.yaml
@@ -0,0 +1,46 @@
+# datagen-config-examples/web_research.yaml
+#
+# Batch data generation config for WebResearchEnv.
+# Generates tool-calling trajectories for multi-step web research tasks.
+#
+# Usage:
+#   python batch_runner.py \
+#     --config datagen-config-examples/web_research.yaml \
+#     --run_name web_research_v1
+
+environment: web-research
+
+# Toolsets available to the agent during data generation
+toolsets:
+  - web
+  - file
+
+# How many parallel workers to use
+num_workers: 4
+
+# Questions per batch
+batch_size: 20
+
+# Total trajectories to generate (comment out to run full dataset)
+max_items: 500
+
+# Model to use for generation (override with --model flag)
+model: openrouter/nousresearch/hermes-3-llama-3.1-405b
+
+# System prompt additions (ephemeral — not saved to trajectories)
+ephemeral_system_prompt: |
+  You are a highly capable research agent. When asked a factual question,
+  always use web_search to find current, accurate information before answering.
+  Cite at least 2 sources. Be concise and accurate.
+
+# Output directory
+output_dir: data/web_research_v1
+
+# Trajectory compression settings (for fitting into training token budgets)
+compression:
+  enabled: true
+  target_max_tokens: 16000
+
+# Eval settings
+eval_every: 100       # Run eval every N trajectories
+eval_size: 25         # Number of held-out questions per eval run
--- a/docs/README.md
+++ b/docs/README.md
@@ -1,7 +0,0 @@
-# Documentation
-
-All documentation has moved to the website:
-
-**📖 [hermes-agent.nousresearch.com/docs](https://hermes-agent.nousresearch.com/docs/)**
-
-The documentation source files live in [`website/docs/`](../website/docs/).
--- a/docs/send_file_integration_map.md
+++ b/docs/send_file_integration_map.md
@@ -1,345 +0,0 @@
-# send_file Integration Map — Hermes Agent Codebase Deep Dive
-
-## 1. environments/tool_context.py — Base64 File Transfer Implementation
-
-### upload_file() (lines 153-205)
- Reads local file as raw bytes, base64-encodes to ASCII string
- Creates parent dirs in sandbox via `self.terminal(f"mkdir -p {parent}")`
- **Chunk size:** 60,000 chars (~60KB per shell command)
- **Small files (<=60KB b64):** Single `printf '%s' '{b64}' | base64 -d > {remote_path}`
- **Large files:** Writes chunks to `/tmp/_hermes_upload.b64` via `printf >> append`, then `base64 -d` to target
- **Error handling:** Checks local file exists; returns `{exit_code, output}`
- **Size limits:** No explicit limit, but shell arg limit ~2MB means chunking is necessary for files >~45KB raw
- **No theoretical max** — but very large files would be slow (many terminal round trips)
-
-### download_file() (lines 234-278)
- Runs `base64 {remote_path}` inside sandbox, captures stdout
- Strips output, base64-decodes to raw bytes
- Writes to host filesystem with parent dir creation
- **Error handling:** Checks exit code, empty output, decode errors
- Returns `{success: bool, bytes: int}` or `{success: false, error: str}`
- **Size limit:** Bounded by terminal output buffer (practical limit ~few MB via base64 terminal output)
-
-### Promotion potential:
- These methods work via `self.terminal()` — they're environment-agnostic
- Could be directly lifted into a new tool that operates on the agent's current sandbox
- For send_file, this `download_file()` pattern is the key: it extracts files from sandbox → host
-
-## 2. tools/environments/base.py — BaseEnvironment Interface
-
-### Current methods:
- `execute(command, cwd, timeout, stdin_data)` → `{output, returncode}`
- `cleanup()` — release resources
- `stop()` — alias for cleanup
- `_prepare_command()` — sudo transformation
- `_build_run_kwargs()` — subprocess kwargs
- `_timeout_result()` — standard timeout dict
-
-### What would need to be added for file transfer:
- **Nothing required at this level.** File transfer can be implemented via `execute()` (base64 over terminal, like ToolContext does) or via environment-specific methods.
- Optional: `upload_file(local_path, remote_path)` and `download_file(remote_path, local_path)` methods could be added to BaseEnvironment for optimized per-backend transfers, but the base64-over-terminal approach already works universally.
-
-## 3. tools/environments/docker.py — Docker Container Details
-
-### Container ID tracking:
- `self._container_id` stored at init from `self._inner.container_id`
- Inner is `minisweagent.environments.docker.DockerEnvironment`
- Container ID is a standard Docker container hash
-
-### docker cp feasibility:
- **YES**, `docker cp` could be used for optimized file transfer:
-  - `docker cp {container_id}:{remote_path} {local_path}` (download)
-  - `docker cp {local_path} {container_id}:{remote_path}` (upload)
- Much faster than base64-over-terminal for large files
- Container ID is directly accessible via `env._container_id` or `env._inner.container_id`
-
-### Volumes mounted:
- **Persistent mode:** Bind mounts at `~/.hermes/sandboxes/docker/{task_id}/workspace` → `/workspace` and `.../home` → `/root`
- **Ephemeral mode:** tmpfs at `/workspace` (10GB), `/home` (1GB), `/root` (1GB)
- **User volumes:** From `config.yaml docker_volumes` (arbitrary `-v` mounts)
- **Security tmpfs:** `/tmp` (512MB), `/var/tmp` (256MB), `/run` (64MB)
-
-### Direct host access for persistent mode:
- If persistent, files at `/workspace/foo.txt` are just `~/.hermes/sandboxes/docker/{task_id}/workspace/foo.txt` on host — no transfer needed!
-
-## 4. tools/environments/ssh.py — SSH Connection Management
-
-### Connection management:
- Uses SSH ControlMaster for persistent connection
- Control socket at `/tmp/hermes-ssh/{user}@{host}:{port}.sock`
- ControlPersist=300 (5 min keepalive)
- BatchMode=yes (non-interactive)
- Stores: `self.host`, `self.user`, `self.port`, `self.key_path`
-
-### SCP/SFTP feasibility:
- **YES**, SCP can piggyback on the ControlMaster socket:
-  - `scp -o ControlPath={socket} {user}@{host}:{remote} {local}` (download)
-  - `scp -o ControlPath={socket} {local} {user}@{host}:{remote}` (upload)
- Same SSH key and connection reuse — zero additional auth
- Would be much faster than base64-over-terminal for large files
-
-## 5. tools/environments/modal.py — Modal Sandbox Filesystem
-
-### Filesystem API exposure:
- **Not directly.** The inner `SwerexModalEnvironment` wraps Modal's sandbox
- The sandbox object is accessible at: `env._inner.deployment._sandbox`
- Modal's Python SDK exposes `sandbox.open()` for file I/O — but only via async API
- Currently only used for `snapshot_filesystem()` during cleanup
- **Could use:** `sandbox.open(path, "rb")` to read files or `sandbox.open(path, "wb")` to write
- **Alternative:** Base64-over-terminal already works via `execute()` — simpler, no SDK dependency
-
-## 6. gateway/platforms/base.py — MEDIA: Tag Flow (Complete)
-
-### extract_media() (lines 587-620):
- **Pattern:** `MEDIA:\S+` — extracts file paths after MEDIA: prefix
- **Voice flag:** `[[audio_as_voice]]` global directive sets `is_voice=True` for all media in message
- Returns `List[Tuple[str, bool]]` (path, is_voice) and cleaned content
-
-### _process_message_background() media routing (lines 752-786):
- After extracting MEDIA tags, routes by file extension:
-  - `.ogg .opus .mp3 .wav .m4a` → `send_voice()`
-  - `.mp4 .mov .avi .mkv .3gp` → `send_video()`
-  - `.jpg .jpeg .png .webp .gif` → `send_image_file()`
-  - **Everything else** → `send_document()`
- This routing already supports arbitrary files!
-
-### send_* method inventory (base class):
- `send(chat_id, content, reply_to, metadata)` — ABSTRACT, text
- `send_image(chat_id, image_url, caption, reply_to)` — URL-based images
- `send_animation(chat_id, animation_url, caption, reply_to)` — GIF animations
- `send_voice(chat_id, audio_path, caption, reply_to)` — voice messages
- `send_video(chat_id, video_path, caption, reply_to)` — video files
- `send_document(chat_id, file_path, caption, file_name, reply_to)` — generic files
- `send_image_file(chat_id, image_path, caption, reply_to)` — local image files
- `send_typing(chat_id)` — typing indicator
- `edit_message(chat_id, message_id, content)` — edit sent messages
-
-### What's missing:
- **Telegram:** No override for `send_document` — falls back to text! (`send_image_file` ✅ added)
- **Discord:** No override for `send_document` — falls back to text! (`send_image_file` ✅ added)
- **Slack:** No override for `send_document` — falls back to text! (`send_image_file` ✅ added)
- **WhatsApp:** Has `send_document` and `send_image_file` via bridge — COMPLETE.
- The base class defaults just send "📎 File: /path" as text — useless for actual file delivery.
-
-## 7. gateway/platforms/telegram.py — Send Method Analysis
-
-### Implemented send methods:
- `send()` — MarkdownV2 text with fallback to plain
- `send_voice()` — `.ogg`/`.opus` as `send_voice()`, others as `send_audio()`
- `send_image()` — URL-based via `send_photo()`
- `send_image_file()` — local file via `send_photo(photo=open(path, 'rb'))` ✅
- `send_animation()` — GIF via `send_animation()`
- `send_typing()` — "typing" chat action
- `edit_message()` — edit text messages
-
-### MISSING:
- **`send_document()` NOT overridden** — Need to add `self._bot.send_document(chat_id, document=open(file_path, 'rb'), ...)`
- **`send_video()` NOT overridden** — Need to add `self._bot.send_video(...)`
-
-## 8. gateway/platforms/discord.py — Send Method Analysis
-
-### Implemented send methods:
- `send()` — text messages with chunking
- `send_voice()` — discord.File attachment
- `send_image()` — downloads URL, creates discord.File attachment
- `send_image_file()` — local file via discord.File attachment ✅
- `send_typing()` — channel.typing()
- `edit_message()` — edit text messages
-
-### MISSING:
- **`send_document()` NOT overridden** — Need to add discord.File attachment
- **`send_video()` NOT overridden** — Need to add discord.File attachment
-
-## 9. gateway/run.py — User File Attachment Handling
-
-### Current attachment flow:
-1. **Telegram photos** (line 509-529): Download via `photo.get_file()` → `cache_image_from_bytes()` → vision auto-analysis
-2. **Telegram voice** (line 532-541): Download → `cache_audio_from_bytes()` → STT transcription
-3. **Telegram audio** (line 542-551): Same pattern
-4. **Telegram documents** (line 553-617): Extension validation against `SUPPORTED_DOCUMENT_TYPES`, 20MB limit, content injection for text files
-5. **Discord attachments** (line 717-751): Content-type detection, image/audio caching, URL fallback for other types
-6. **Gateway run.py** (lines 818-883): Auto-analyzes images with vision, transcribes audio, enriches document messages with context notes
-
-### Key insight: Files are always cached to host filesystem first, then processed. The agent sees local file paths.
-
-## 10. tools/terminal_tool.py — Terminal Tool & Environment Interaction
-
-### How it manages environments:
- Global dict `_active_environments: Dict[str, Any]` keyed by task_id
- Per-task creation locks prevent duplicate sandbox creation
- Auto-cleanup thread kills idle environments after `TERMINAL_LIFETIME_SECONDS`
- `_get_env_config()` reads all TERMINAL_* env vars for backend selection
- `_create_environment()` factory creates the right backend type
-
-### Could send_file piggyback?
- **YES.** send_file needs access to the same environment to extract files from sandboxes.
- It can reuse `_active_environments[task_id]` to get the environment, then:
-  - Docker: Use `docker cp` via `env._container_id`
-  - SSH: Use `scp` via `env.control_socket`
-  - Local: Just read the file directly
-  - Modal: Use base64-over-terminal via `env.execute()`
- The file_tools.py module already does this with `ShellFileOperations` — read_file/write_file/search/patch all share the same env instance.
-
-## 11. tools/tts_tool.py — Working Example of File Delivery
-
-### Flow:
-1. Generate audio file to `~/.hermes/audio_cache/tts_TIMESTAMP.{ogg,mp3}`
-2. Return JSON with `media_tag: "MEDIA:/path/to/file"`
-3. For Telegram voice: prepend `[[audio_as_voice]]` directive
-4. The LLM includes the MEDIA tag in its response text
-5. `BasePlatformAdapter._process_message_background()` calls `extract_media()` to find the tag
-6. Routes by extension → `send_voice()` for audio files
-7. Platform adapter sends the file natively
-
-### Key pattern: Tool saves file to host → returns MEDIA: path → LLM echoes it → gateway extracts → platform delivers
-
-## 12. tools/image_generation_tool.py — Working Example of Image Delivery
-
-### Flow:
-1. Call FAL.ai API → get image URL
-2. Return JSON with `image: "https://fal.media/..."` URL
-3. The LLM includes the URL in markdown: `![description](URL)`
-4. `BasePlatformAdapter.extract_images()` finds `![alt](url)` patterns
-5. Routes through `send_image()` (URL) or `send_animation()` (GIF)
-6. Platform downloads and sends natively
-
-### Key difference from TTS: Images are URL-based, not local files. The gateway downloads at send time.
-
---
-
-# INTEGRATION MAP: Where send_file Hooks In
-
-## Architecture Decision: MEDIA: Tag Protocol vs. New Tool
-
-The MEDIA: tag protocol is already the established pattern for file delivery. Two options:
-
-### Option A: Pure MEDIA: Tag (Minimal Change)
- No new tool needed
- Agent downloads file from sandbox to host using terminal (base64)
- Saves to known location (e.g., `~/.hermes/file_cache/`)
- Includes `MEDIA:/path` in response text
- Existing routing in `_process_message_background()` handles delivery
- **Problem:** Agent has to manually do base64 dance + know about MEDIA: convention
-
-### Option B: Dedicated send_file Tool (Recommended)
- New tool that the agent calls with `(file_path, caption?)`
- Tool handles the sandbox → host extraction automatically
- Returns MEDIA: tag that gets routed through existing pipeline
- Much cleaner agent experience
-
-## Implementation Plan for Option B
-
-### Files to CREATE:
-
-1. **`tools/send_file_tool.py`** — The new tool
-   - Accepts: `file_path` (path in sandbox), `caption` (optional)
-   - Detects environment backend from `_active_environments`
-   - Extracts file from sandbox:
-     - **local:** `shutil.copy()` or direct path
-     - **docker:** `docker cp {container_id}:{path} {local_cache}/` 
-     - **ssh:** `scp -o ControlPath=... {user}@{host}:{path} {local_cache}/`
-     - **modal:** base64-over-terminal via `env.execute("base64 {path}")`
-   - Saves to `~/.hermes/file_cache/{uuid}_{filename}`
-   - Returns: `MEDIA:/cached/path` in response for gateway to pick up
-   - Register with `registry.register(name="send_file", toolset="file", ...)`
-
-### Files to MODIFY:
-
-2. **`gateway/platforms/telegram.py`** — Add missing send methods:
-   ```python
-   async def send_document(self, chat_id, file_path, caption=None, file_name=None, reply_to=None):
-       with open(file_path, "rb") as f:
-           msg = await self._bot.send_document(
-               chat_id=int(chat_id), document=f,
-               caption=caption, filename=file_name or os.path.basename(file_path))
-       return SendResult(success=True, message_id=str(msg.message_id))
-   
-   async def send_image_file(self, chat_id, image_path, caption=None, reply_to=None):
-       with open(image_path, "rb") as f:
-           msg = await self._bot.send_photo(chat_id=int(chat_id), photo=f, caption=caption)
-       return SendResult(success=True, message_id=str(msg.message_id))
-   
-   async def send_video(self, chat_id, video_path, caption=None, reply_to=None):
-       with open(video_path, "rb") as f:
-           msg = await self._bot.send_video(chat_id=int(chat_id), video=f, caption=caption)
-       return SendResult(success=True, message_id=str(msg.message_id))
-   ```
-
-3. **`gateway/platforms/discord.py`** — Add missing send methods:
-   ```python
-   async def send_document(self, chat_id, file_path, caption=None, file_name=None, reply_to=None):
-       channel = self._client.get_channel(int(chat_id)) or await self._client.fetch_channel(int(chat_id))
-       with open(file_path, "rb") as f:
-           file = discord.File(io.BytesIO(f.read()), filename=file_name or os.path.basename(file_path))
-           msg = await channel.send(content=caption, file=file)
-       return SendResult(success=True, message_id=str(msg.id))
-   
-   async def send_image_file(self, chat_id, image_path, caption=None, reply_to=None):
-       # Same pattern as send_document with image filename
-   
-   async def send_video(self, chat_id, video_path, caption=None, reply_to=None):
-       # Same pattern, discord renders video attachments inline
-   ```
-
-4. **`toolsets.py`** — Add `"send_file"` to `_HERMES_CORE_TOOLS` list
-
-5. **`agent/prompt_builder.py`** — Update platform hints to mention send_file tool
-
-### Code that can be REUSED (zero rewrite):
-
- `BasePlatformAdapter.extract_media()` — Already extracts MEDIA: tags
- `BasePlatformAdapter._process_message_background()` — Already routes by extension
- `ToolContext.download_file()` — Base64-over-terminal extraction pattern
- `tools/terminal_tool.py` _active_environments dict — Environment access
- `tools/registry.py` — Tool registration infrastructure
- `gateway/platforms/base.py` send_document/send_image_file/send_video signatures — Already defined
-
-### Code that needs to be WRITTEN from scratch:
-
-1. `tools/send_file_tool.py` (~150 lines):
-   - File extraction from each environment backend type
-   - Local file cache management
-   - Registry registration
-   
-2. Telegram `send_document` + `send_image_file` + `send_video` overrides (~40 lines)
-3. Discord `send_document` + `send_image_file` + `send_video` overrides (~50 lines)
-
-### Total effort: ~240 lines of new code, ~5 lines of config changes
-
-## Key Environment-Specific Extract Strategies
-
-| Backend    | Extract Method                 | Speed    | Complexity |
-|------------|-------------------------------|----------|------------|
-| local      | shutil.copy / direct path     | Instant  | None       |
-| docker     | `docker cp container:path .`  | Fast     | Low        |
-| docker+vol | Direct host path access       | Instant  | None       |
-| ssh        | `scp -o ControlPath=...`      | Fast     | Low        |
-| modal      | base64-over-terminal          | Moderate | Medium     |
-| singularity| Direct path (overlay mount)   | Fast     | Low        |
-
-## Data Flow Summary
-
-```
-Agent calls send_file(file_path="/workspace/output.pdf", caption="Here's the report")
-    │
-    ▼
-send_file_tool.py:
-    1. Get environment from _active_environments[task_id]
-    2. Detect backend type (docker/ssh/modal/local)
-    3. Extract file to ~/.hermes/file_cache/{uuid}_{filename}
-    4. Return: '{"success": true, "media_tag": "MEDIA:/home/user/.hermes/file_cache/abc123_output.pdf"}'
-    │
-    ▼
-LLM includes MEDIA: tag in its response text
-    │
-    ▼
-BasePlatformAdapter._process_message_background():
-    1. extract_media(response) → finds MEDIA:/path
-    2. Checks extension: .pdf → send_document()
-    3. Calls platform-specific send_document(chat_id, file_path, caption)
-    │
-    ▼
-TelegramAdapter.send_document() / DiscordAdapter.send_document():
-    Opens file, sends via platform API as native document attachment
-    User receives downloadable file in chat
-```
--- a/docs/skins/example-skin.yaml
+++ b/docs/skins/example-skin.yaml
@@ -0,0 +1,89 @@
+# ============================================================================
+# Hermes Agent — Example Skin Template
+# ============================================================================
+#
+# Copy this file to ~/.hermes/skins/<name>.yaml to create a custom skin.
+# All fields are optional — missing values inherit from the default skin.
+# Activate with: /skin <name>  or  display.skin: <name> in config.yaml
+#
+# See hermes_cli/skin_engine.py for the full schema reference.
+# ============================================================================
+
+# Required: unique skin name (used in /skin command and config)
+name: example
+description: An example custom skin — copy and modify this template
+
+# ── Colors ──────────────────────────────────────────────────────────────────
+# Hex color values for Rich markup. These control the CLI's visual palette.
+colors:
+  # Banner panel (the startup welcome box)
+  banner_border: "#CD7F32"        # Panel border
+  banner_title: "#FFD700"         # Panel title text
+  banner_accent: "#FFBF00"        # Section headers (Available Tools, Skills, etc.)
+  banner_dim: "#B8860B"           # Dim/muted text (separators, model info)
+  banner_text: "#FFF8DC"          # Body text (tool names, skill names)
+
+  # UI elements
+  ui_accent: "#FFBF00"            # General accent color
+  ui_label: "#4dd0e1"             # Labels
+  ui_ok: "#4caf50"                # Success indicators
+  ui_error: "#ef5350"             # Error indicators
+  ui_warn: "#ffa726"              # Warning indicators
+
+  # Input area
+  prompt: "#FFF8DC"               # Prompt text color
+  input_rule: "#CD7F32"           # Horizontal rule around input
+
+  # Response box
+  response_border: "#FFD700"      # Response box border (ANSI color)
+
+  # Session display
+  session_label: "#DAA520"        # Session label
+  session_border: "#8B8682"       # Session ID dim color
+
+# ── Spinner ─────────────────────────────────────────────────────────────────
+# Customize the animated spinner shown during API calls and tool execution.
+spinner:
+  # Faces shown while waiting for the API response
+  waiting_faces:
+    - "(｡◕‿◕｡)"
+    - "(◕‿◕✿)"
+    - "٩(◕‿◕｡)۶"
+
+  # Faces shown during extended thinking/reasoning
+  thinking_faces:
+    - "(｡•́︿•̀｡)"
+    - "(◔_◔)"
+    - "(¬‿¬)"
+
+  # Verbs used in spinner messages (e.g., "pondering your request...")
+  thinking_verbs:
+    - "pondering"
+    - "contemplating"
+    - "musing"
+    - "ruminating"
+
+  # Optional: left/right decorations around the spinner
+  # Each entry is a [left, right] pair. Omit entirely for no wings.
+  # wings:
+  #   - ["⟪⚔", "⚔⟫"]
+  #   - ["⟪▲", "▲⟫"]
+
+# ── Branding ────────────────────────────────────────────────────────────────
+# Text strings used throughout the CLI interface.
+branding:
+  agent_name: "Hermes Agent"          # Banner title, about display
+  welcome: "Welcome! Type your message or /help for commands."
+  goodbye: "Goodbye! ⚕"              # Exit message
+  response_label: " ⚕ Hermes "       # Response box header label
+  prompt_symbol: "❯ "                 # Input prompt symbol
+  help_header: "(^_^)? Available Commands"  # /help header text
+
+# ── Tool Output ─────────────────────────────────────────────────────────────
+# Character used as the prefix for tool output lines.
+# Default is "┊" (thin dotted vertical line). Some alternatives:
+#   "╎" (light triple dash vertical)
+#   "▏" (left one-eighth block)
+#   "│" (box drawing light vertical)
+#   "┃" (box drawing heavy vertical)
+tool_prefix: "┊"
--- a/environments/benchmarks/terminalbench_2/default.yaml
+++ b/environments/benchmarks/terminalbench_2/default.yaml
@@ -29,6 +29,10 @@ env:
  wandb_name: "terminal-bench-2"
  ensure_scores_are_not_same: false
  data_dir_to_save_evals: "environments/benchmarks/evals/terminal-bench-2"
+  # CRITICAL: Limit concurrent Modal sandbox creations to avoid deadlocks.
+  # Modal's blocking calls (App.lookup, etc.) deadlock when too many sandboxes
+  # are created simultaneously inside thread pool workers via asyncio.run().
+  max_concurrent_tasks: 8

 openai:
  base_url: "https://openrouter.ai/api/v1"
--- a/environments/benchmarks/terminalbench_2/terminalbench2_env.py
+++ b/environments/benchmarks/terminalbench_2/terminalbench2_env.py
@@ -118,6 +118,15 @@ class TerminalBench2EvalConfig(HermesAgentEnvConfig):
        "Tasks exceeding this are scored as FAIL. Default 30 minutes.",
    )

+    # --- Concurrency control ---
+    max_concurrent_tasks: int = Field(
+        default=8,
+        description="Maximum number of tasks to run concurrently. "
+        "Limits concurrent Modal sandbox creations to avoid async/threading deadlocks. "
+        "Modal has internal limits and creating too many sandboxes simultaneously "
+        "causes blocking calls to deadlock inside the thread pool.",
+    )
+

 # Tasks that cannot run properly on Modal and are excluded from scoring.
 MODAL_INCOMPATIBLE_TASKS = {
@@ -430,7 +439,7 @@ class TerminalBench2EvalEnv(HermesAgentBaseEnv):
                }

            # --- 2. Register per-task Modal image override ---
-            register_task_env_overrides(task_id, {"modal_image": modal_image})
+            register_task_env_overrides(task_id, {"modal_image": modal_image, "cwd": "/app"})
            logger.info(
                "Task %s: registered image override for task_id %s",
                task_name, task_id[:8],
@@ -733,12 +742,23 @@ class TerminalBench2EvalEnv(HermesAgentBaseEnv):
        print(f"  Tool thread pool: {self.config.tool_pool_size}")
        print(f"  Terminal timeout: {self.config.terminal_timeout}s/cmd")
        print(f"  Terminal lifetime: {self.config.terminal_lifetime}s (auto: task_timeout + 120)")
+        print(f"  Max concurrent tasks: {self.config.max_concurrent_tasks}")
        print(f"{'='*60}\n")

+        # Semaphore to limit concurrent Modal sandbox creations.
+        # Without this, all 86 tasks fire simultaneously, each creating a Modal
+        # sandbox via asyncio.run() inside a thread pool worker. Modal's blocking
+        # calls (App.lookup, etc.) deadlock when too many are created at once.
+        semaphore = asyncio.Semaphore(self.config.max_concurrent_tasks)
+
+        async def _eval_with_semaphore(item):
+            async with semaphore:
+                return await self._eval_with_timeout(item)
+
        # Fire all tasks with wall-clock timeout, track live accuracy on the bar
        total_tasks = len(self.all_eval_items)
        eval_tasks = [
-            asyncio.ensure_future(self._eval_with_timeout(item))
+            asyncio.ensure_future(_eval_with_semaphore(item))
            for item in self.all_eval_items
        ]

--- a/environments/web_research_env.py
+++ b/environments/web_research_env.py
@@ -0,0 +1,718 @@
+"""
+WebResearchEnv — RL Environment for Multi-Step Web Research
+============================================================
+
+Trains models to do accurate, efficient, multi-source web research.
+
+Reward signals:
+  - Answer correctness  (LLM judge, 0.0–1.0)
+  - Source diversity    (used ≥2 distinct domains)
+  - Efficiency          (penalizes excessive tool calls)
+  - Tool usage          (bonus for actually using web tools)
+
+Dataset: FRAMES benchmark (Google, 2024) — multi-hop factual questions
+  HuggingFace: google/frames-benchmark
+  Fallback:    built-in sample questions (no HF token needed)
+
+Usage:
+    # Phase 1 (OpenAI-compatible server)
+    python environments/web_research_env.py serve \\
+        --openai.base_url http://localhost:8000/v1 \\
+        --openai.model_name YourModel \\
+        --openai.server_type openai
+
+    # Process mode (offline data generation)
+    python environments/web_research_env.py process \\
+        --env.data_path_to_save_groups data/web_research.jsonl
+
+    # Standalone eval
+    python environments/web_research_env.py evaluate \\
+        --openai.base_url http://localhost:8000/v1 \\
+        --openai.model_name YourModel
+
+Built by: github.com/jackx707
+Inspired by: GroceryMind — production Hermes agent doing live web research
+             across German grocery stores (firecrawl + hermes-agent)
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import logging
+import os
+import random
+import re
+import sys
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Tuple
+from urllib.parse import urlparse
+
+from pydantic import Field
+
+# Ensure hermes-agent root is on path
+_repo_root = Path(__file__).resolve().parent.parent
+if str(_repo_root) not in sys.path:
+    sys.path.insert(0, str(_repo_root))
+
+# ---------------------------------------------------------------------------
+# Optional HuggingFace datasets import
+# ---------------------------------------------------------------------------
+try:
+    from datasets import load_dataset
+    HF_AVAILABLE = True
+except ImportError:
+    HF_AVAILABLE = False
+
+from atroposlib.envs.base import ScoredDataGroup
+from atroposlib.envs.server_handling.server_manager import APIServerConfig
+from atroposlib.type_definitions import Item
+
+from environments.hermes_base_env import HermesAgentBaseEnv, HermesAgentEnvConfig
+from environments.agent_loop import AgentResult
+from environments.tool_context import ToolContext
+
+logger = logging.getLogger(__name__)
+
+# ---------------------------------------------------------------------------
+# Fallback sample dataset (used when HuggingFace is unavailable)
+# Multi-hop questions requiring real web search to answer.
+# ---------------------------------------------------------------------------
+SAMPLE_QUESTIONS = [
+    {
+        "question": "What is the current population of the capital city of the country that won the 2022 FIFA World Cup?",
+        "answer": "Buenos Aires has approximately 3 million people in the city proper, or around 15 million in the greater metro area.",
+        "difficulty": "medium",
+        "hops": 2,
+    },
+    {
+        "question": "Who is the CEO of the company that makes the most widely used open-source container orchestration platform?",
+        "answer": "The Linux Foundation oversees Kubernetes. CNCF (Cloud Native Computing Foundation) is the specific body — it does not have a traditional CEO but has an executive director.",
+        "difficulty": "medium",
+        "hops": 2,
+    },
+    {
+        "question": "What programming language was used to write the original version of the web framework used by Instagram?",
+        "answer": "Django, which Instagram was built on, is written in Python.",
+        "difficulty": "easy",
+        "hops": 2,
+    },
+    {
+        "question": "In what year was the university founded where the inventor of the World Wide Web currently holds a professorship?",
+        "answer": "Tim Berners-Lee holds a professorship at MIT (founded 1861) and the University of Southampton (founded 1952).",
+        "difficulty": "hard",
+        "hops": 3,
+    },
+    {
+        "question": "What is the latest stable version of the programming language that ranks #1 on the TIOBE index as of this year?",
+        "answer": "Python is currently #1 on TIOBE. The latest stable version should be verified via the official python.org site.",
+        "difficulty": "medium",
+        "hops": 2,
+    },
+    {
+        "question": "How many employees does the parent company of Instagram have?",
+        "answer": "Meta Platforms (parent of Instagram) employs approximately 70,000+ people as of recent reports.",
+        "difficulty": "medium",
+        "hops": 2,
+    },
+    {
+        "question": "What is the current interest rate set by the central bank of the country where the Eiffel Tower is located?",
+        "answer": "The European Central Bank sets rates for France/eurozone. The current rate should be verified — it has changed frequently in 2023-2025.",
+        "difficulty": "hard",
+        "hops": 2,
+    },
+    {
+        "question": "Which company acquired the startup founded by the creator of Oculus VR?",
+        "answer": "Palmer Luckey founded Oculus VR, which was acquired by Facebook (now Meta). He later founded Anduril Industries.",
+        "difficulty": "medium",
+        "hops": 2,
+    },
+    {
+        "question": "What is the market cap of the company that owns the most popular search engine in Russia?",
+        "answer": "Yandex (now split into separate entities after 2024 restructuring). Current market cap should be verified via financial sources.",
+        "difficulty": "hard",
+        "hops": 2,
+    },
+    {
+        "question": "What was the GDP growth rate of the country that hosted the most recent Summer Olympics?",
+        "answer": "Paris, France hosted the 2024 Summer Olympics. France's recent GDP growth should be verified via World Bank or IMF data.",
+        "difficulty": "hard",
+        "hops": 2,
+    },
+]
+
+
+# ---------------------------------------------------------------------------
+# Configuration
+# ---------------------------------------------------------------------------
+
+class WebResearchEnvConfig(HermesAgentEnvConfig):
+    """Configuration for the web research RL environment."""
+
+    # Reward weights
+    correctness_weight: float = Field(
+        default=0.6,
+        description="Weight for answer correctness in reward (LLM judge score).",
+    )
+    tool_usage_weight: float = Field(
+        default=0.2,
+        description="Weight for tool usage signal (did the model actually use web tools?).",
+    )
+    efficiency_weight: float = Field(
+        default=0.2,
+        description="Weight for efficiency signal (penalizes excessive tool calls).",
+    )
+    diversity_bonus: float = Field(
+        default=0.1,
+        description="Bonus reward for citing ≥2 distinct domains.",
+    )
+
+    # Efficiency thresholds
+    efficient_max_calls: int = Field(
+        default=5,
+        description="Maximum tool calls before efficiency penalty begins.",
+    )
+    heavy_penalty_calls: int = Field(
+        default=10,
+        description="Tool call count where efficiency penalty steepens.",
+    )
+
+    # Eval
+    eval_size: int = Field(
+        default=20,
+        description="Number of held-out items for evaluation.",
+    )
+    eval_split_ratio: float = Field(
+        default=0.1,
+        description="Fraction of dataset to hold out for evaluation (0.0–1.0).",
+    )
+
+    # Dataset
+    dataset_name: str = Field(
+        default="google/frames-benchmark",
+        description="HuggingFace dataset name for research questions.",
+    )
+
+
+# ---------------------------------------------------------------------------
+# Environment
+# ---------------------------------------------------------------------------
+
+class WebResearchEnv(HermesAgentBaseEnv):
+    """
+    RL environment for training multi-step web research skills.
+
+    The model is given a factual question requiring 2-3 hops of web research
+    and must use web_search / web_extract tools to find and synthesize the answer.
+
+    Reward is multi-signal:
+      60% — answer correctness (LLM judge)
+      20% — tool usage (did the model actually search the web?)
+      20% — efficiency (penalizes >5 tool calls)
+
+    Bonus +0.1 for source diversity (≥2 distinct domains cited).
+    """
+
+    name = "web-research"
+    env_config_cls = WebResearchEnvConfig
+
+    # Default toolsets for this environment — web + file for saving notes
+    default_toolsets = ["web", "file"]
+
+    @classmethod
+    def config_init(cls) -> Tuple[WebResearchEnvConfig, List[APIServerConfig]]:
+        """Default configuration for the web research environment."""
+        env_config = WebResearchEnvConfig(
+            enabled_toolsets=["web", "file"],
+            max_agent_turns=15,
+            agent_temperature=1.0,
+            system_prompt=(
+                "You are a highly capable research agent. When asked a factual question, "
+                "always use web_search to find current, accurate information before answering. "
+                "Cite at least 2 sources. Be concise and accurate."
+            ),
+            group_size=4,
+            total_steps=1000,
+            steps_per_eval=100,
+            use_wandb=True,
+            wandb_name="web-research",
+        )
+
+        server_configs = [
+            APIServerConfig(
+                base_url="https://openrouter.ai/api/v1",
+                model_name="anthropic/claude-sonnet-4.5",
+                server_type="openai",
+                api_key=os.getenv("OPENROUTER_API_KEY", ""),
+                health_check=False,
+            )
+        ]
+
+        return env_config, server_configs
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self._items: list[dict] = []
+        self._eval_items: list[dict] = []
+        self._index: int = 0
+
+        # Metrics tracking for wandb
+        self._reward_buffer: list[float] = []
+        self._correctness_buffer: list[float] = []
+        self._tool_usage_buffer: list[float] = []
+        self._efficiency_buffer: list[float] = []
+        self._diversity_buffer: list[float] = []
+
+    # ------------------------------------------------------------------
+    # 1. Setup — load dataset
+    # ------------------------------------------------------------------
+
+    async def setup(self) -> None:
+        """Load the FRAMES benchmark or fall back to built-in samples."""
+        if HF_AVAILABLE:
+            try:
+                logger.info("Loading FRAMES benchmark from HuggingFace...")
+                ds = load_dataset(self.config.dataset_name, split="test")
+                self._items = [
+                    {
+                        "question": row["Prompt"],
+                        "answer": row["Answer"],
+                        "difficulty": row.get("reasoning_types", "unknown"),
+                        "hops": 2,
+                    }
+                    for row in ds
+                ]
+                # Hold out for eval
+                eval_size = max(
+                    self.config.eval_size,
+                    int(len(self._items) * self.config.eval_split_ratio),
+                )
+                random.shuffle(self._items)
+                self._eval_items = self._items[:eval_size]
+                self._items = self._items[eval_size:]
+                logger.info(
+                    f"Loaded {len(self._items)} train / {len(self._eval_items)} eval items "
+                    f"from FRAMES benchmark."
+                )
+                return
+            except Exception as e:
+                logger.warning(f"Could not load FRAMES from HuggingFace: {e}. Using built-in samples.")
+
+        # Fallback
+        random.shuffle(SAMPLE_QUESTIONS)
+        split = max(1, len(SAMPLE_QUESTIONS) * 8 // 10)
+        self._items = SAMPLE_QUESTIONS[:split]
+        self._eval_items = SAMPLE_QUESTIONS[split:]
+        logger.info(
+            f"Using built-in sample dataset: {len(self._items)} train / "
+            f"{len(self._eval_items)} eval items."
+        )
+
+    # ------------------------------------------------------------------
+    # 2. get_next_item — return the next question
+    # ------------------------------------------------------------------
+
+    async def get_next_item(self) -> dict:
+        """Return the next item, cycling through the dataset."""
+        if not self._items:
+            raise RuntimeError("Dataset is empty. Did you call setup()?")
+        item = self._items[self._index % len(self._items)]
+        self._index += 1
+        return item
+
+    # ------------------------------------------------------------------
+    # 3. format_prompt — build the user-facing prompt
+    # ------------------------------------------------------------------
+
+    def format_prompt(self, item: dict) -> str:
+        """Format the research question as a task prompt."""
+        return (
+            f"Research the following question thoroughly using web search. "
+            f"You MUST search the web to find current, accurate information — "
+            f"do not rely solely on your training data.\n\n"
+            f"Question: {item['question']}\n\n"
+            f"Requirements:\n"
+            f"- Use web_search and/or web_extract tools to find information\n"
+            f"- Search at least 2 different sources\n"
+            f"- Provide a concise, accurate answer (2-4 sentences)\n"
+            f"- Cite the sources you used"
+        )
+
+    # ------------------------------------------------------------------
+    # 4. compute_reward — multi-signal scoring
+    # ------------------------------------------------------------------
+
+    async def compute_reward(
+        self,
+        item: dict,
+        result: AgentResult,
+        ctx: ToolContext,
+    ) -> float:
+        """
+        Multi-signal reward function:
+
+          correctness_weight * correctness  — LLM judge comparing answer to ground truth
+          tool_usage_weight  * tool_used    — binary: did the model use web tools?
+          efficiency_weight  * efficiency   — penalizes wasteful tool usage
+          + diversity_bonus                 — source diversity (≥2 distinct domains)
+        """
+        # Extract final response from messages (last assistant message with content)
+        final_response = ""
+        tools_used: list[str] = []
+        for msg in reversed(result.messages):
+            if msg.get("role") == "assistant" and msg.get("content") and not final_response:
+                final_response = msg["content"]
+            # Collect tool names from tool call messages
+            if msg.get("role") == "assistant" and msg.get("tool_calls"):
+                for tc in msg["tool_calls"]:
+                    fn = tc.get("function", {}) if isinstance(tc, dict) else {}
+                    name = fn.get("name", "")
+                    if name:
+                        tools_used.append(name)
+        tool_call_count: int = result.turns_used or len(tools_used)
+
+        cfg = self.config
+
+        # ---- Signal 1: Answer correctness (LLM judge) ----------------
+        correctness = await self._llm_judge(
+            question=item["question"],
+            expected=item["answer"],
+            model_answer=final_response,
+        )
+
+        # ---- Signal 2: Web tool usage --------------------------------
+        web_tools = {"web_search", "web_extract", "search", "firecrawl"}
+        tool_used = 1.0 if any(t in web_tools for t in tools_used) else 0.0
+
+        # ---- Signal 3: Efficiency ------------------------------------
+        if tool_call_count <= cfg.efficient_max_calls:
+            efficiency = 1.0
+        elif tool_call_count <= cfg.heavy_penalty_calls:
+            efficiency = 1.0 - (tool_call_count - cfg.efficient_max_calls) * 0.08
+        else:
+            efficiency = max(0.0, 1.0 - (tool_call_count - cfg.efficient_max_calls) * 0.12)
+
+        # ---- Bonus: Source diversity ---------------------------------
+        domains = self._extract_domains(final_response)
+        diversity = cfg.diversity_bonus if len(domains) >= 2 else 0.0
+
+        # ---- Combine ------------------------------------------------
+        reward = (
+            cfg.correctness_weight * correctness
+            + cfg.tool_usage_weight * tool_used
+            + cfg.efficiency_weight * efficiency
+            + diversity
+        )
+        reward = min(1.0, max(0.0, reward))  # clamp to [0, 1]
+
+        # Track for wandb
+        self._reward_buffer.append(reward)
+        self._correctness_buffer.append(correctness)
+        self._tool_usage_buffer.append(tool_used)
+        self._efficiency_buffer.append(efficiency)
+        self._diversity_buffer.append(diversity)
+
+        logger.debug(
+            f"Reward breakdown — correctness={correctness:.2f}, "
+            f"tool_used={tool_used:.1f}, efficiency={efficiency:.2f}, "
+            f"diversity={diversity:.1f} → total={reward:.3f}"
+        )
+
+        return reward
+
+    # ------------------------------------------------------------------
+    # 5. evaluate — run on held-out eval split
+    # ------------------------------------------------------------------
+
+    async def evaluate(self, *args, **kwargs) -> None:
+        """Run evaluation on the held-out split using the full agent loop with tools.
+
+        Each eval item runs through the same agent loop as training —
+        the model can use web_search, web_extract, etc. to research answers.
+        This measures actual agentic research capability, not just knowledge.
+        """
+        import time
+        import uuid
+        from environments.agent_loop import HermesAgentLoop
+        from environments.tool_context import ToolContext
+
+        items = self._eval_items
+        if not items:
+            logger.warning("No eval items available.")
+            return
+
+        eval_size = min(self.config.eval_size, len(items))
+        eval_items = items[:eval_size]
+
+        logger.info(f"Running eval on {len(eval_items)} questions (with agent loop + tools)...")
+        start_time = time.time()
+        samples = []
+
+        # Resolve tools once for all eval items
+        tools, valid_names = self._resolve_tools_for_group()
+
+        for i, item in enumerate(eval_items):
+            task_id = str(uuid.uuid4())
+            logger.info(f"Eval [{i+1}/{len(eval_items)}]: {item['question'][:80]}...")
+
+            try:
+                # Build messages
+                messages: List[Dict[str, Any]] = []
+                if self.config.system_prompt:
+                    messages.append({"role": "system", "content": self.config.system_prompt})
+                messages.append({"role": "user", "content": self.format_prompt(item)})
+
+                # Run the full agent loop with tools
+                agent = HermesAgentLoop(
+                    server=self.server,
+                    tool_schemas=tools,
+                    valid_tool_names=valid_names,
+                    max_turns=self.config.max_agent_turns,
+                    task_id=task_id,
+                    temperature=0.0,  # Deterministic for eval
+                    max_tokens=self.config.max_token_length,
+                    extra_body=self.config.extra_body,
+                )
+                result = await agent.run(messages)
+
+                # Extract final response and tool usage from messages
+                final_response = ""
+                tool_call_count = 0
+                for msg in reversed(result.messages):
+                    if msg.get("role") == "assistant" and msg.get("content") and not final_response:
+                        final_response = msg["content"]
+                    if msg.get("role") == "assistant" and msg.get("tool_calls"):
+                        tool_call_count += len(msg["tool_calls"])
+
+                # Compute reward (includes LLM judge for correctness)
+                # Temporarily save buffer lengths so we can extract the
+                # correctness score without calling judge twice, and avoid
+                # polluting training metric buffers with eval data.
+                buf_len = len(self._correctness_buffer)
+                ctx = ToolContext(task_id)
+                try:
+                    reward = await self.compute_reward(item, result, ctx)
+                finally:
+                    ctx.cleanup()
+
+                # Extract correctness from the buffer (compute_reward appended it)
+                # then remove eval entries from training buffers
+                correctness = (
+                    self._correctness_buffer[buf_len]
+                    if len(self._correctness_buffer) > buf_len
+                    else 0.0
+                )
+                # Roll back buffers to avoid polluting training metrics
+                for buf in (
+                    self._reward_buffer, self._correctness_buffer,
+                    self._tool_usage_buffer, self._efficiency_buffer,
+                    self._diversity_buffer,
+                ):
+                    if len(buf) > buf_len:
+                        buf.pop()
+
+                samples.append({
+                    "prompt": item["question"],
+                    "response": final_response[:500],
+                    "expected": item["answer"],
+                    "correctness": correctness,
+                    "reward": reward,
+                    "tool_calls": tool_call_count,
+                    "turns": result.turns_used,
+                })
+
+                logger.info(
+                    f"  → correctness={correctness:.2f}, reward={reward:.3f}, "
+                    f"tools={tool_call_count}, turns={result.turns_used}"
+                )
+
+            except Exception as e:
+                logger.error(f"Eval error on item: {e}")
+                samples.append({
+                    "prompt": item["question"],
+                    "response": f"ERROR: {e}",
+                    "expected": item["answer"],
+                    "correctness": 0.0,
+                    "reward": 0.0,
+                    "tool_calls": 0,
+                    "turns": 0,
+                })
+
+        end_time = time.time()
+
+        # Compute aggregate metrics
+        correctness_scores = [s["correctness"] for s in samples]
+        rewards = [s["reward"] for s in samples]
+        tool_counts = [s["tool_calls"] for s in samples]
+        n = len(samples)
+
+        eval_metrics = {
+            "eval/mean_correctness": sum(correctness_scores) / n if n else 0.0,
+            "eval/mean_reward": sum(rewards) / n if n else 0.0,
+            "eval/mean_tool_calls": sum(tool_counts) / n if n else 0.0,
+            "eval/tool_usage_rate": sum(1 for t in tool_counts if t > 0) / n if n else 0.0,
+            "eval/n_items": n,
+        }
+
+        logger.info(
+            f"Eval complete — correctness={eval_metrics['eval/mean_correctness']:.3f}, "
+            f"reward={eval_metrics['eval/mean_reward']:.3f}, "
+            f"tool_usage={eval_metrics['eval/tool_usage_rate']:.0%}"
+        )
+
+        await self.evaluate_log(
+            metrics=eval_metrics,
+            samples=samples,
+            start_time=start_time,
+            end_time=end_time,
+        )
+
+    # ------------------------------------------------------------------
+    # 6. wandb_log — custom metrics
+    # ------------------------------------------------------------------
+
+    async def wandb_log(self, wandb_metrics: Optional[Dict] = None) -> None:
+        """Log reward breakdown metrics to wandb."""
+        if wandb_metrics is None:
+            wandb_metrics = {}
+
+        if self._reward_buffer:
+            n = len(self._reward_buffer)
+            wandb_metrics["train/mean_reward"] = sum(self._reward_buffer) / n
+            wandb_metrics["train/mean_correctness"] = sum(self._correctness_buffer) / n
+            wandb_metrics["train/mean_tool_usage"] = sum(self._tool_usage_buffer) / n
+            wandb_metrics["train/mean_efficiency"] = sum(self._efficiency_buffer) / n
+            wandb_metrics["train/mean_diversity"] = sum(self._diversity_buffer) / n
+            wandb_metrics["train/total_rollouts"] = n
+
+            # Accuracy buckets
+            wandb_metrics["train/correct_rate"] = (
+                sum(1 for c in self._correctness_buffer if c >= 0.7) / n
+            )
+            wandb_metrics["train/tool_usage_rate"] = (
+                sum(1 for t in self._tool_usage_buffer if t > 0) / n
+            )
+
+            # Clear buffers
+            self._reward_buffer.clear()
+            self._correctness_buffer.clear()
+            self._tool_usage_buffer.clear()
+            self._efficiency_buffer.clear()
+            self._diversity_buffer.clear()
+
+        await super().wandb_log(wandb_metrics)
+
+    # ------------------------------------------------------------------
+    # Private helpers
+    # ------------------------------------------------------------------
+
+    async def _llm_judge(
+        self,
+        question: str,
+        expected: str,
+        model_answer: str,
+    ) -> float:
+        """
+        Use the server's LLM to judge answer correctness.
+        Falls back to keyword heuristic if LLM call fails.
+        """
+        if not model_answer or not model_answer.strip():
+            return 0.0
+
+        judge_prompt = (
+            "You are an impartial judge evaluating the quality of an AI research answer.\n\n"
+            f"Question: {question}\n\n"
+            f"Reference answer: {expected}\n\n"
+            f"Model answer: {model_answer}\n\n"
+            "Score the model answer on a scale from 0.0 to 1.0 where:\n"
+            "  1.0 = fully correct and complete\n"
+            "  0.7 = mostly correct with minor gaps\n"
+            "  0.4 = partially correct\n"
+            "  0.1 = mentions relevant topic but wrong or very incomplete\n"
+            "  0.0 = completely wrong or no answer\n\n"
+            "Consider: factual accuracy, completeness, and relevance.\n"
+            'Respond with ONLY a JSON object: {"score": <float>, "reason": "<one sentence>"}'
+        )
+
+        try:
+            response = await self.server.chat_completion(
+                messages=[{"role": "user", "content": judge_prompt}],
+                n=1,
+                max_tokens=150,
+                temperature=0.0,
+                split="eval",
+            )
+            text = response.choices[0].message.content if response.choices else ""
+            parsed = self._parse_judge_json(text)
+            if parsed is not None:
+                return float(parsed)
+        except Exception as e:
+            logger.debug(f"LLM judge failed: {e}. Using heuristic.")
+
+        return self._heuristic_score(expected, model_answer)
+
+    @staticmethod
+    def _parse_judge_json(text: str) -> Optional[float]:
+        """Extract the score float from LLM judge JSON response."""
+        try:
+            clean = re.sub(r"```(?:json)?|```", "", text).strip()
+            data = json.loads(clean)
+            score = float(data.get("score", -1))
+            if 0.0 <= score <= 1.0:
+                return score
+        except Exception:
+            match = re.search(r'"score"\s*:\s*([0-9.]+)', text)
+            if match:
+                score = float(match.group(1))
+                if 0.0 <= score <= 1.0:
+                    return score
+        return None
+
+    @staticmethod
+    def _heuristic_score(expected: str, model_answer: str) -> float:
+        """Lightweight keyword overlap score as fallback."""
+        stopwords = {
+            "the", "a", "an", "is", "are", "was", "were", "of", "in", "on",
+            "at", "to", "for", "with", "and", "or", "but", "it", "its",
+            "this", "that", "as", "by", "from", "be", "has", "have", "had",
+        }
+
+        def tokenize(text: str) -> set:
+            tokens = re.findall(r'\b\w+\b', text.lower())
+            return {t for t in tokens if t not in stopwords and len(t) > 2}
+
+        expected_tokens = tokenize(expected)
+        answer_tokens = tokenize(model_answer)
+
+        if not expected_tokens:
+            return 0.5
+
+        overlap = len(expected_tokens & answer_tokens)
+        union = len(expected_tokens | answer_tokens)
+
+        jaccard = overlap / union if union > 0 else 0.0
+        recall = overlap / len(expected_tokens)
+        return min(1.0, 0.4 * jaccard + 0.6 * recall)
+
+    @staticmethod
+    def _extract_domains(text: str) -> set:
+        """Extract unique domains from URLs cited in the response."""
+        urls = re.findall(r'https?://[^\s\)>\]"\']+', text)
+        domains = set()
+        for url in urls:
+            try:
+                parsed = urlparse(url)
+                domain = parsed.netloc.lower().lstrip("www.")
+                if domain:
+                    domains.add(domain)
+            except Exception:
+                pass
+        return domains
+
+
+# ---------------------------------------------------------------------------
+# Entry point
+# ---------------------------------------------------------------------------
+
+if __name__ == "__main__":
+    WebResearchEnv.cli()
--- a/gateway/channel_directory.py
+++ b/gateway/channel_directory.py
@@ -40,8 +40,8 @@ def build_channel_directory(adapters: Dict[Any, Any]) -> Dict[str, Any]:
        except Exception as e:
            logger.warning("Channel directory: failed to build %s: %s", platform.value, e)

-    # Telegram & WhatsApp can't enumerate chats -- pull from session history
-    for plat_name in ("telegram", "whatsapp"):
+    # Telegram, WhatsApp & Signal can't enumerate chats -- pull from session history
+    for plat_name in ("telegram", "whatsapp", "signal"):
        if plat_name not in platforms:
            platforms[plat_name] = _build_from_sessions(plat_name)

@@ -52,7 +52,7 @@ def build_channel_directory(adapters: Dict[Any, Any]) -> Dict[str, Any]:

    try:
        DIRECTORY_PATH.parent.mkdir(parents=True, exist_ok=True)
-        with open(DIRECTORY_PATH, "w") as f:
+        with open(DIRECTORY_PATH, "w", encoding="utf-8") as f:
            json.dump(directory, f, indent=2, ensure_ascii=False)
    except Exception as e:
        logger.warning("Channel directory: failed to write: %s", e)
@@ -115,7 +115,7 @@ def _build_from_sessions(platform_name: str) -> List[Dict[str, str]]:

    entries = []
    try:
-        with open(sessions_path) as f:
+        with open(sessions_path, encoding="utf-8") as f:
            data = json.load(f)

        seen_ids = set()
@@ -147,7 +147,7 @@ def load_directory() -> Dict[str, Any]:
    if not DIRECTORY_PATH.exists():
        return {"updated_at": None, "platforms": {}}
    try:
-        with open(DIRECTORY_PATH) as f:
+        with open(DIRECTORY_PATH, encoding="utf-8") as f:
            return json.load(f)
    except Exception:
        return {"updated_at": None, "platforms": {}}
--- a/gateway/config.py
+++ b/gateway/config.py
@@ -26,6 +26,7 @@ class Platform(Enum):
    DISCORD = "discord"
    WHATSAPP = "whatsapp"
    SLACK = "slack"
+    SIGNAL = "signal"
    HOMEASSISTANT = "homeassistant"


@@ -155,7 +156,16 @@ class GatewayConfig:
        """Return list of platforms that are enabled and configured."""
        connected = []
        for platform, config in self.platforms.items():
-            if config.enabled and (config.token or config.api_key):
+            if not config.enabled:
+                continue
+            # Platforms that use token/api_key auth
+            if config.token or config.api_key:
+                connected.append(platform)
+            # WhatsApp uses enabled flag only (bridge handles auth)
+            elif platform == Platform.WHATSAPP:
+                connected.append(platform)
+            # Signal uses extra dict for config (http_url + account)
+            elif platform == Platform.SIGNAL and config.extra.get("http_url"):
                connected.append(platform)
        return connected
    
@@ -260,7 +270,7 @@ def load_gateway_config() -> GatewayConfig:
    gateway_config_path = Path.home() / ".hermes" / "gateway.json"
    if gateway_config_path.exists():
        try:
-            with open(gateway_config_path, "r") as f:
+            with open(gateway_config_path, "r", encoding="utf-8") as f:
                data = json.load(f)
                config = GatewayConfig.from_dict(data)
        except Exception as e:
@@ -273,7 +283,7 @@ def load_gateway_config() -> GatewayConfig:
        import yaml
        config_yaml_path = Path.home() / ".hermes" / "config.yaml"
        if config_yaml_path.exists():
-            with open(config_yaml_path) as f:
+            with open(config_yaml_path, encoding="utf-8") as f:
                yaml_cfg = yaml.safe_load(f) or {}
            sr = yaml_cfg.get("session_reset")
            if sr and isinstance(sr, dict):
@@ -379,6 +389,26 @@ def _apply_env_overrides(config: GatewayConfig) -> None:
                name=os.getenv("SLACK_HOME_CHANNEL_NAME", ""),
            )
    
+    # Signal
+    signal_url = os.getenv("SIGNAL_HTTP_URL")
+    signal_account = os.getenv("SIGNAL_ACCOUNT")
+    if signal_url and signal_account:
+        if Platform.SIGNAL not in config.platforms:
+            config.platforms[Platform.SIGNAL] = PlatformConfig()
+        config.platforms[Platform.SIGNAL].enabled = True
+        config.platforms[Platform.SIGNAL].extra.update({
+            "http_url": signal_url,
+            "account": signal_account,
+            "ignore_stories": os.getenv("SIGNAL_IGNORE_STORIES", "true").lower() in ("true", "1", "yes"),
+        })
+        signal_home = os.getenv("SIGNAL_HOME_CHANNEL")
+        if signal_home:
+            config.platforms[Platform.SIGNAL].home_channel = HomeChannel(
+                platform=Platform.SIGNAL,
+                chat_id=signal_home,
+                name=os.getenv("SIGNAL_HOME_CHANNEL_NAME", "Home"),
+            )
+
    # Home Assistant
    hass_token = os.getenv("HASS_TOKEN")
    if hass_token:
@@ -411,5 +441,5 @@ def save_gateway_config(config: GatewayConfig) -> None:
    gateway_config_path = Path.home() / ".hermes" / "gateway.json"
    gateway_config_path.parent.mkdir(parents=True, exist_ok=True)
    
-    with open(gateway_config_path, "w") as f:
+    with open(gateway_config_path, "w", encoding="utf-8") as f:
        json.dump(config.to_dict(), f, indent=2)
--- a/gateway/mirror.py
+++ b/gateway/mirror.py
@@ -73,7 +73,7 @@ def _find_session_id(platform: str, chat_id: str) -> Optional[str]:
        return None

    try:
-        with open(_SESSIONS_INDEX) as f:
+        with open(_SESSIONS_INDEX, encoding="utf-8") as f:
            data = json.load(f)
    except Exception:
        return None
@@ -103,7 +103,7 @@ def _append_to_jsonl(session_id: str, message: dict) -> None:
    """Append a message to the JSONL transcript file."""
    transcript_path = _SESSIONS_DIR / f"{session_id}.jsonl"
    try:
-        with open(transcript_path, "a") as f:
+        with open(transcript_path, "a", encoding="utf-8") as f:
            f.write(json.dumps(message, ensure_ascii=False) + "\n")
    except Exception as e:
        logger.debug("Mirror JSONL write failed: %s", e)
@@ -111,6 +111,7 @@ def _append_to_jsonl(session_id: str, message: dict) -> None:

 def _append_to_sqlite(session_id: str, message: dict) -> None:
    """Append a message to the SQLite session database."""
+    db = None
    try:
        from hermes_state import SessionDB
        db = SessionDB()
@@ -121,3 +122,6 @@ def _append_to_sqlite(session_id: str, message: dict) -> None:
        )
    except Exception as e:
        logger.debug("Mirror SQLite write failed: %s", e)
+    finally:
+        if db is not None:
+            db.close()
--- a/gateway/platforms/ADDING_A_PLATFORM.md
+++ b/gateway/platforms/ADDING_A_PLATFORM.md
@@ -0,0 +1,313 @@
+# Adding a New Messaging Platform
+
+Checklist for integrating a new messaging platform into the Hermes gateway.
+Use this as a reference when building a new adapter — every item here is a
+real integration point that exists in the codebase. Missing any of them will
+cause broken functionality, missing features, or inconsistent behavior.
+
+---
+
+## 1. Core Adapter (`gateway/platforms/<platform>.py`)
+
+The adapter is a subclass of `BasePlatformAdapter` from `gateway/platforms/base.py`.
+
+### Required methods
+
+| Method | Purpose |
+|--------|---------|
+| `__init__(self, config)` | Parse config, init state. Call `super().__init__(config, Platform.YOUR_PLATFORM)` |
+| `connect() -> bool` | Connect to the platform, start listeners. Return True on success |
+| `disconnect()` | Stop listeners, close connections, cancel tasks |
+| `send(chat_id, text, ...) -> SendResult` | Send a text message |
+| `send_typing(chat_id)` | Send typing indicator |
+| `send_image(chat_id, image_url, caption) -> SendResult` | Send an image |
+| `get_chat_info(chat_id) -> dict` | Return `{name, type, chat_id}` for a chat |
+
+### Optional methods (have default stubs in base)
+
+| Method | Purpose |
+|--------|---------|
+| `send_document(chat_id, path, caption)` | Send a file attachment |
+| `send_voice(chat_id, path)` | Send a voice message |
+| `send_video(chat_id, path, caption)` | Send a video |
+| `send_animation(chat_id, path, caption)` | Send a GIF/animation |
+| `send_image_file(chat_id, path, caption)` | Send image from local file |
+
+### Required function
+
+```python
+def check_<platform>_requirements() -> bool:
+    """Check if this platform's dependencies are available."""
+```
+
+### Key patterns to follow
+
+- Use `self.build_source(...)` to construct `SessionSource` objects
+- Call `self.handle_message(event)` to dispatch inbound messages to the gateway
+- Use `MessageEvent`, `MessageType`, `SendResult` from base
+- Use `cache_image_from_bytes`, `cache_audio_from_bytes`, `cache_document_from_bytes` for attachments
+- Filter self-messages (prevent reply loops)
+- Filter sync/echo messages if the platform has them
+- Redact sensitive identifiers (phone numbers, tokens) in all log output
+- Implement reconnection with exponential backoff + jitter for streaming connections
+- Set `MAX_MESSAGE_LENGTH` if the platform has message size limits
+
+---
+
+## 2. Platform Enum (`gateway/config.py`)
+
+Add the platform to the `Platform` enum:
+
+```python
+class Platform(Enum):
+    ...
+    YOUR_PLATFORM = "your_platform"
+```
+
+Add env var loading in `_apply_env_overrides()`:
+
+```python
+# Your Platform
+your_token = os.getenv("YOUR_PLATFORM_TOKEN")
+if your_token:
+    if Platform.YOUR_PLATFORM not in config.platforms:
+        config.platforms[Platform.YOUR_PLATFORM] = PlatformConfig()
+    config.platforms[Platform.YOUR_PLATFORM].enabled = True
+    config.platforms[Platform.YOUR_PLATFORM].token = your_token
+```
+
+Update `get_connected_platforms()` if your platform doesn't use token/api_key
+(e.g., WhatsApp uses `enabled` flag, Signal uses `extra` dict).
+
+---
+
+## 3. Adapter Factory (`gateway/run.py`)
+
+Add to `_create_adapter()`:
+
+```python
+elif platform == Platform.YOUR_PLATFORM:
+    from gateway.platforms.your_platform import YourAdapter, check_your_requirements
+    if not check_your_requirements():
+        logger.warning("Your Platform: dependencies not met")
+        return None
+    return YourAdapter(config)
+```
+
+---
+
+## 4. Authorization Maps (`gateway/run.py`)
+
+Add to BOTH dicts in `_is_user_authorized()`:
+
+```python
+platform_env_map = {
+    ...
+    Platform.YOUR_PLATFORM: "YOUR_PLATFORM_ALLOWED_USERS",
+}
+platform_allow_all_map = {
+    ...
+    Platform.YOUR_PLATFORM: "YOUR_PLATFORM_ALLOW_ALL_USERS",
+}
+```
+
+---
+
+## 5. Session Source (`gateway/session.py`)
+
+If your platform needs extra identity fields (e.g., Signal's UUID alongside
+phone number), add them to the `SessionSource` dataclass with `Optional` defaults,
+and update `to_dict()`, `from_dict()`, and `build_source()` in base.py.
+
+---
+
+## 6. System Prompt Hints (`agent/prompt_builder.py`)
+
+Add a `PLATFORM_HINTS` entry so the agent knows what platform it's on:
+
+```python
+PLATFORM_HINTS = {
+    ...
+    "your_platform": (
+        "You are on Your Platform. "
+        "Describe formatting capabilities, media support, etc."
+    ),
+}
+```
+
+Without this, the agent won't know it's on your platform and may use
+inappropriate formatting (e.g., markdown on platforms that don't render it).
+
+---
+
+## 7. Toolset (`toolsets.py`)
+
+Add a named toolset for your platform:
+
+```python
+"hermes-your-platform": {
+    "description": "Your Platform bot toolset",
+    "tools": _HERMES_CORE_TOOLS,
+    "includes": []
+},
+```
+
+And add it to the `hermes-gateway` composite:
+
+```python
+"hermes-gateway": {
+    "includes": [..., "hermes-your-platform"]
+}
+```
+
+---
+
+## 8. Cron Delivery (`cron/scheduler.py`)
+
+Add to `platform_map` in `_deliver_result()`:
+
+```python
+platform_map = {
+    ...
+    "your_platform": Platform.YOUR_PLATFORM,
+}
+```
+
+Without this, `schedule_cronjob(deliver="your_platform")` silently fails.
+
+---
+
+## 9. Send Message Tool (`tools/send_message_tool.py`)
+
+Add to `platform_map` in `send_message_tool()`:
+
+```python
+platform_map = {
+    ...
+    "your_platform": Platform.YOUR_PLATFORM,
+}
+```
+
+Add routing in `_send_to_platform()`:
+
+```python
+elif platform == Platform.YOUR_PLATFORM:
+    return await _send_your_platform(pconfig, chat_id, message)
+```
+
+Implement `_send_your_platform()` — a standalone async function that sends
+a single message without requiring the full adapter (for use by cron jobs
+and the send_message tool outside the gateway process).
+
+Update the tool schema `target` description to include your platform example.
+
+---
+
+## 10. Cronjob Tool Schema (`tools/cronjob_tools.py`)
+
+Update the `deliver` parameter description and docstring to mention your
+platform as a delivery option.
+
+---
+
+## 11. Channel Directory (`gateway/channel_directory.py`)
+
+If your platform can't enumerate chats (most can't), add it to the
+session-based discovery list:
+
+```python
+for plat_name in ("telegram", "whatsapp", "signal", "your_platform"):
+```
+
+---
+
+## 12. Status Display (`hermes_cli/status.py`)
+
+Add to the `platforms` dict in the Messaging Platforms section:
+
+```python
+platforms = {
+    ...
+    "Your Platform": ("YOUR_PLATFORM_TOKEN", "YOUR_PLATFORM_HOME_CHANNEL"),
+}
+```
+
+---
+
+## 13. Gateway Setup Wizard (`hermes_cli/gateway.py`)
+
+Add to the `_PLATFORMS` list:
+
+```python
+{
+    "key": "your_platform",
+    "label": "Your Platform",
+    "emoji": "📱",
+    "token_var": "YOUR_PLATFORM_TOKEN",
+    "setup_instructions": [...],
+    "vars": [...],
+}
+```
+
+If your platform needs custom setup logic (connectivity testing, QR codes,
+policy choices), add a `_setup_your_platform()` function and route to it
+in the platform selection switch.
+
+Update `_platform_status()` if your platform's "configured" check differs
+from the standard `bool(get_env_value(token_var))`.
+
+---
+
+## 14. Phone/ID Redaction (`agent/redact.py`)
+
+If your platform uses sensitive identifiers (phone numbers, etc.), add a
+regex pattern and redaction function to `agent/redact.py`. This ensures
+identifiers are masked in ALL log output, not just your adapter's logs.
+
+---
+
+## 15. Documentation
+
+| File | What to update |
+|------|---------------|
+| `README.md` | Platform list in feature table + documentation table |
+| `AGENTS.md` | Gateway description + env var config section |
+| `website/docs/user-guide/messaging/<platform>.md` | **NEW** — Full setup guide (see existing platform docs for template) |
+| `website/docs/user-guide/messaging/index.md` | Architecture diagram, toolset table, security examples, Next Steps links |
+| `website/docs/reference/environment-variables.md` | All env vars for the platform |
+
+---
+
+## 16. Tests (`tests/gateway/test_<platform>.py`)
+
+Recommended test coverage:
+
+- Platform enum exists with correct value
+- Config loading from env vars via `_apply_env_overrides`
+- Adapter init (config parsing, allowlist handling, default values)
+- Helper functions (redaction, parsing, file type detection)
+- Session source round-trip (to_dict → from_dict)
+- Authorization integration (platform in allowlist maps)
+- Send message tool routing (platform in platform_map)
+
+Optional but valuable:
+- Async tests for message handling flow (mock the platform API)
+- SSE/WebSocket reconnection logic
+- Attachment processing
+- Group message filtering
+
+---
+
+## Quick Verification
+
+After implementing everything, verify with:
+
+```bash
+# All tests pass
+python -m pytest tests/ -q
+
+# Grep for your platform name to find any missed integration points
+grep -r "telegram\|discord\|whatsapp\|slack" gateway/ tools/ agent/ cron/ hermes_cli/ toolsets.py \
+  --include="*.py" -l | sort -u
+# Check each file in the output — if it mentions other platforms but not yours, you missed it
+```
--- a/gateway/platforms/base.py
+++ b/gateway/platforms/base.py
@@ -252,6 +252,7 @@ def cleanup_document_cache(max_age_hours: int = 24) -> int:
 class MessageType(Enum):
    """Types of incoming messages."""
    TEXT = "text"
+    LOCATION = "location"
    PHOTO = "photo"
    VIDEO = "video"
    AUDIO = "audio"
@@ -412,11 +413,12 @@ class BasePlatformAdapter(ABC):
        """
        return SendResult(success=False, error="Not supported")

-    async def send_typing(self, chat_id: str) -> None:
+    async def send_typing(self, chat_id: str, metadata=None) -> None:
        """
        Send a typing indicator.
        
        Override in subclasses if the platform supports it.
+        metadata: optional dict with platform-specific context (e.g. thread_id for Slack).
        """
        pass
    
@@ -619,7 +621,7 @@ class BasePlatformAdapter(ABC):
        
        return media, cleaned
    
-    async def _keep_typing(self, chat_id: str, interval: float = 2.0) -> None:
+    async def _keep_typing(self, chat_id: str, interval: float = 2.0, metadata=None) -> None:
        """
        Continuously send typing indicator until cancelled.
        
@@ -628,7 +630,7 @@ class BasePlatformAdapter(ABC):
        """
        try:
            while True:
-                await self.send_typing(chat_id)
+                await self.send_typing(chat_id, metadata=metadata)
                await asyncio.sleep(interval)
        except asyncio.CancelledError:
            pass  # Normal cancellation when handler completes
@@ -686,7 +688,8 @@ class BasePlatformAdapter(ABC):
        self._active_sessions[session_key] = interrupt_event
        
        # Start continuous typing indicator (refreshes every 2 seconds)
-        typing_task = asyncio.create_task(self._keep_typing(event.source.chat_id))
+        _thread_metadata = {"thread_id": event.source.thread_id} if event.source.thread_id else None
+        typing_task = asyncio.create_task(self._keep_typing(event.source.chat_id, metadata=_thread_metadata))
        
        try:
            # Call the handler (this can take a while with tool calls)
@@ -710,7 +713,8 @@ class BasePlatformAdapter(ABC):
                    result = await self.send(
                        chat_id=event.source.chat_id,
                        content=text_content,
-                        reply_to=event.message_id
+                        reply_to=event.message_id,
+                        metadata=_thread_metadata,
                    )
                    
                    # Log send failures (don't raise - user already saw tool progress)
@@ -720,7 +724,8 @@ class BasePlatformAdapter(ABC):
                        fallback_result = await self.send(
                            chat_id=event.source.chat_id,
                            content=f"(Response formatting failed, plain text:)\n\n{text_content[:3500]}",
-                            reply_to=event.message_id
+                            reply_to=event.message_id,
+                            metadata=_thread_metadata,
                        )
                        if not fallback_result.success:
                            print(f"[{self.name}] Fallback send also failed: {fallback_result.error}")
@@ -742,12 +747,14 @@ class BasePlatformAdapter(ABC):
                                chat_id=event.source.chat_id,
                                animation_url=image_url,
                                caption=alt_text if alt_text else None,
+                                metadata=_thread_metadata,
                            )
                        else:
                            img_result = await self.send_image(
                                chat_id=event.source.chat_id,
                                image_url=image_url,
                                caption=alt_text if alt_text else None,
+                                metadata=_thread_metadata,
                            )
                        if not img_result.success:
                            logger.error("[%s] Failed to send image: %s", self.name, img_result.error)
@@ -768,21 +775,25 @@ class BasePlatformAdapter(ABC):
                            media_result = await self.send_voice(
                                chat_id=event.source.chat_id,
                                audio_path=media_path,
+                                metadata=_thread_metadata,
                            )
                        elif ext in _VIDEO_EXTS:
                            media_result = await self.send_video(
                                chat_id=event.source.chat_id,
                                video_path=media_path,
+                                metadata=_thread_metadata,
                            )
                        elif ext in _IMAGE_EXTS:
                            media_result = await self.send_image_file(
                                chat_id=event.source.chat_id,
                                image_path=media_path,
+                                metadata=_thread_metadata,
                            )
                        else:
                            media_result = await self.send_document(
                                chat_id=event.source.chat_id,
                                file_path=media_path,
+                                metadata=_thread_metadata,
                            )

                        if not media_result.success:
@@ -838,6 +849,8 @@ class BasePlatformAdapter(ABC):
        user_name: Optional[str] = None,
        thread_id: Optional[str] = None,
        chat_topic: Optional[str] = None,
+        user_id_alt: Optional[str] = None,
+        chat_id_alt: Optional[str] = None,
    ) -> SessionSource:
        """Helper to build a SessionSource for this platform."""
        # Normalize empty topic to None
@@ -852,6 +865,8 @@ class BasePlatformAdapter(ABC):
            user_name=user_name,
            thread_id=str(thread_id) if thread_id else None,
            chat_topic=chat_topic.strip() if chat_topic else None,
+            user_id_alt=user_id_alt,
+            chat_id_alt=chat_id_alt,
        )
    
    @abstractmethod
--- a/gateway/platforms/discord.py
+++ b/gateway/platforms/discord.py
@@ -359,7 +359,7 @@ class DiscordAdapter(BasePlatformAdapter):
            print(f"[{self.name}] Failed to send image attachment, falling back to URL: {e}")
            return await super().send_image(chat_id, image_url, caption, reply_to)
    
-    async def send_typing(self, chat_id: str) -> None:
+    async def send_typing(self, chat_id: str, metadata=None) -> None:
        """Send typing indicator."""
        if self._client:
            try:
--- a/gateway/platforms/homeassistant.py
+++ b/gateway/platforms/homeassistant.py
@@ -419,7 +419,7 @@ class HomeAssistantAdapter(BasePlatformAdapter):
        except Exception as e:
            return SendResult(success=False, error=str(e))

-    async def send_typing(self, chat_id: str) -> None:
+    async def send_typing(self, chat_id: str, metadata=None) -> None:
        """No typing indicator for Home Assistant."""
        pass

--- a/gateway/platforms/signal.py
+++ b/gateway/platforms/signal.py
@@ -0,0 +1,727 @@
+"""Signal messenger platform adapter.
+
+Connects to a signal-cli daemon running in HTTP mode.
+Inbound messages arrive via SSE (Server-Sent Events) streaming.
+Outbound messages and actions use JSON-RPC 2.0 over HTTP.
+
+Based on PR #268 by ibhagwan, rebuilt with bug fixes.
+
+Requires:
+  - signal-cli installed and running: signal-cli daemon --http 127.0.0.1:8080
+  - SIGNAL_HTTP_URL and SIGNAL_ACCOUNT environment variables set
+"""
+
+import asyncio
+import base64
+import json
+import logging
+import os
+import random
+import re
+import time
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Dict, List, Optional, Any
+from urllib.parse import unquote
+
+import httpx
+
+from gateway.config import Platform, PlatformConfig
+from gateway.platforms.base import (
+    BasePlatformAdapter,
+    MessageEvent,
+    MessageType,
+    SendResult,
+    cache_image_from_bytes,
+    cache_audio_from_bytes,
+    cache_document_from_bytes,
+    cache_image_from_url,
+)
+
+logger = logging.getLogger(__name__)
+
+# ---------------------------------------------------------------------------
+# Constants
+# ---------------------------------------------------------------------------
+SIGNAL_MAX_ATTACHMENT_SIZE = 100 * 1024 * 1024  # 100 MB
+MAX_MESSAGE_LENGTH = 8000  # Signal message size limit
+TYPING_INTERVAL = 8.0  # seconds between typing indicator refreshes
+SSE_RETRY_DELAY_INITIAL = 2.0
+SSE_RETRY_DELAY_MAX = 60.0
+HEALTH_CHECK_INTERVAL = 30.0  # seconds between health checks
+HEALTH_CHECK_STALE_THRESHOLD = 120.0  # seconds without SSE activity before concern
+
+# E.164 phone number pattern for redaction
+_PHONE_RE = re.compile(r"\+[1-9]\d{6,14}")
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+def _redact_phone(phone: str) -> str:
+    """Redact a phone number for logging: +15551234567 -> +155****4567."""
+    if not phone:
+        return "<none>"
+    if len(phone) <= 8:
+        return phone[:2] + "****" + phone[-2:] if len(phone) > 4 else "****"
+    return phone[:4] + "****" + phone[-4:]
+
+
+def _parse_comma_list(value: str) -> List[str]:
+    """Split a comma-separated string into a list, stripping whitespace."""
+    return [v.strip() for v in value.split(",") if v.strip()]
+
+
+def _guess_extension(data: bytes) -> str:
+    """Guess file extension from magic bytes."""
+    if data[:4] == b"\x89PNG":
+        return ".png"
+    if data[:2] == b"\xff\xd8":
+        return ".jpg"
+    if data[:4] == b"GIF8":
+        return ".gif"
+    if len(data) >= 12 and data[:4] == b"RIFF" and data[8:12] == b"WEBP":
+        return ".webp"
+    if data[:4] == b"%PDF":
+        return ".pdf"
+    if len(data) >= 8 and data[4:8] == b"ftyp":
+        return ".mp4"
+    if data[:4] == b"OggS":
+        return ".ogg"
+    if len(data) >= 2 and data[0] == 0xFF and (data[1] & 0xE0) == 0xE0:
+        return ".mp3"
+    if data[:2] == b"PK":
+        return ".zip"
+    return ".bin"
+
+
+def _is_image_ext(ext: str) -> bool:
+    return ext.lower() in (".jpg", ".jpeg", ".png", ".gif", ".webp")
+
+
+def _is_audio_ext(ext: str) -> bool:
+    return ext.lower() in (".mp3", ".wav", ".ogg", ".m4a", ".aac")
+
+
+_EXT_TO_MIME = {
+    ".jpg": "image/jpeg", ".jpeg": "image/jpeg", ".png": "image/png",
+    ".gif": "image/gif", ".webp": "image/webp",
+    ".ogg": "audio/ogg", ".mp3": "audio/mpeg", ".wav": "audio/wav",
+    ".m4a": "audio/mp4", ".aac": "audio/aac",
+    ".mp4": "video/mp4", ".pdf": "application/pdf", ".zip": "application/zip",
+}
+
+
+def _ext_to_mime(ext: str) -> str:
+    """Map file extension to MIME type."""
+    return _EXT_TO_MIME.get(ext.lower(), "application/octet-stream")
+
+
+def _render_mentions(text: str, mentions: list) -> str:
+    """Replace Signal mention placeholders (\\uFFFC) with readable @identifiers.
+
+    Signal encodes @mentions as the Unicode object replacement character
+    with out-of-band metadata containing the mentioned user's UUID/number.
+    """
+    if not mentions or "\uFFFC" not in text:
+        return text
+    # Sort mentions by start position (reverse) to replace from end to start
+    # so indices don't shift as we replace
+    sorted_mentions = sorted(mentions, key=lambda m: m.get("start", 0), reverse=True)
+    for mention in sorted_mentions:
+        start = mention.get("start", 0)
+        length = mention.get("length", 1)
+        # Use the mention's number or UUID as the replacement
+        identifier = mention.get("number") or mention.get("uuid") or "user"
+        replacement = f"@{identifier}"
+        text = text[:start] + replacement + text[start + length:]
+    return text
+
+
+def check_signal_requirements() -> bool:
+    """Check if Signal is configured (has URL and account)."""
+    return bool(os.getenv("SIGNAL_HTTP_URL") and os.getenv("SIGNAL_ACCOUNT"))
+
+
+# ---------------------------------------------------------------------------
+# Signal Adapter
+# ---------------------------------------------------------------------------
+
+class SignalAdapter(BasePlatformAdapter):
+    """Signal messenger adapter using signal-cli HTTP daemon."""
+
+    platform = Platform.SIGNAL
+
+    def __init__(self, config: PlatformConfig):
+        super().__init__(config, Platform.SIGNAL)
+
+        extra = config.extra or {}
+        self.http_url = extra.get("http_url", "http://127.0.0.1:8080").rstrip("/")
+        self.account = extra.get("account", "")
+        self.ignore_stories = extra.get("ignore_stories", True)
+
+        # Parse allowlists — group policy is derived from presence of group allowlist
+        group_allowed_str = os.getenv("SIGNAL_GROUP_ALLOWED_USERS", "")
+        self.group_allow_from = set(_parse_comma_list(group_allowed_str))
+
+        # HTTP client
+        self.client: Optional[httpx.AsyncClient] = None
+
+        # Background tasks
+        self._sse_task: Optional[asyncio.Task] = None
+        self._health_monitor_task: Optional[asyncio.Task] = None
+        self._typing_tasks: Dict[str, asyncio.Task] = {}
+        self._running = False
+        self._last_sse_activity = 0.0
+        self._sse_response: Optional[httpx.Response] = None
+
+        # Normalize account for self-message filtering
+        self._account_normalized = self.account.strip()
+
+        logger.info("Signal adapter initialized: url=%s account=%s groups=%s",
+                     self.http_url, _redact_phone(self.account),
+                     "enabled" if self.group_allow_from else "disabled")
+
+    # ------------------------------------------------------------------
+    # Lifecycle
+    # ------------------------------------------------------------------
+
+    async def connect(self) -> bool:
+        """Connect to signal-cli daemon and start SSE listener."""
+        if not self.http_url or not self.account:
+            logger.error("Signal: SIGNAL_HTTP_URL and SIGNAL_ACCOUNT are required")
+            return False
+
+        self.client = httpx.AsyncClient(timeout=30.0)
+
+        # Health check — verify signal-cli daemon is reachable
+        try:
+            resp = await self.client.get(f"{self.http_url}/api/v1/check", timeout=10.0)
+            if resp.status_code != 200:
+                logger.error("Signal: health check failed (status %d)", resp.status_code)
+                return False
+        except Exception as e:
+            logger.error("Signal: cannot reach signal-cli at %s: %s", self.http_url, e)
+            return False
+
+        self._running = True
+        self._last_sse_activity = time.time()
+        self._sse_task = asyncio.create_task(self._sse_listener())
+        self._health_monitor_task = asyncio.create_task(self._health_monitor())
+
+        logger.info("Signal: connected to %s", self.http_url)
+        return True
+
+    async def disconnect(self) -> None:
+        """Stop SSE listener and clean up."""
+        self._running = False
+
+        if self._sse_task:
+            self._sse_task.cancel()
+            try:
+                await self._sse_task
+            except asyncio.CancelledError:
+                pass
+
+        if self._health_monitor_task:
+            self._health_monitor_task.cancel()
+            try:
+                await self._health_monitor_task
+            except asyncio.CancelledError:
+                pass
+
+        # Cancel all typing tasks
+        for task in self._typing_tasks.values():
+            task.cancel()
+        self._typing_tasks.clear()
+
+        if self.client:
+            await self.client.aclose()
+            self.client = None
+
+        logger.info("Signal: disconnected")
+
+    # ------------------------------------------------------------------
+    # SSE Streaming (inbound messages)
+    # ------------------------------------------------------------------
+
+    async def _sse_listener(self) -> None:
+        """Listen for SSE events from signal-cli daemon."""
+        url = f"{self.http_url}/api/v1/events?account={self.account}"
+        backoff = SSE_RETRY_DELAY_INITIAL
+
+        while self._running:
+            try:
+                logger.debug("Signal SSE: connecting to %s", url)
+                async with self.client.stream(
+                    "GET", url,
+                    headers={"Accept": "text/event-stream"},
+                    timeout=None,
+                ) as response:
+                    self._sse_response = response
+                    backoff = SSE_RETRY_DELAY_INITIAL  # Reset on successful connection
+                    self._last_sse_activity = time.time()
+                    logger.info("Signal SSE: connected")
+
+                    buffer = ""
+                    async for chunk in response.aiter_text():
+                        if not self._running:
+                            break
+                        buffer += chunk
+                        while "\n" in buffer:
+                            line, buffer = buffer.split("\n", 1)
+                            line = line.strip()
+                            if not line:
+                                continue
+                            # Parse SSE data lines
+                            if line.startswith("data:"):
+                                data_str = line[5:].strip()
+                                if not data_str:
+                                    continue
+                                self._last_sse_activity = time.time()
+                                try:
+                                    data = json.loads(data_str)
+                                    await self._handle_envelope(data)
+                                except json.JSONDecodeError:
+                                    logger.debug("Signal SSE: invalid JSON: %s", data_str[:100])
+                                except Exception:
+                                    logger.exception("Signal SSE: error handling event")
+
+            except asyncio.CancelledError:
+                break
+            except httpx.HTTPError as e:
+                if self._running:
+                    logger.warning("Signal SSE: HTTP error: %s (reconnecting in %.0fs)", e, backoff)
+            except Exception as e:
+                if self._running:
+                    logger.warning("Signal SSE: error: %s (reconnecting in %.0fs)", e, backoff)
+
+            if self._running:
+                # Add 20% jitter to prevent thundering herd on reconnection
+                jitter = backoff * 0.2 * random.random()
+                await asyncio.sleep(backoff + jitter)
+                backoff = min(backoff * 2, SSE_RETRY_DELAY_MAX)
+
+        self._sse_response = None
+
+    # ------------------------------------------------------------------
+    # Health Monitor
+    # ------------------------------------------------------------------
+
+    async def _health_monitor(self) -> None:
+        """Monitor SSE connection health and force reconnect if stale."""
+        while self._running:
+            await asyncio.sleep(HEALTH_CHECK_INTERVAL)
+            if not self._running:
+                break
+
+            elapsed = time.time() - self._last_sse_activity
+            if elapsed > HEALTH_CHECK_STALE_THRESHOLD:
+                logger.warning("Signal: SSE idle for %.0fs, checking daemon health", elapsed)
+                try:
+                    resp = await self.client.get(
+                        f"{self.http_url}/api/v1/check", timeout=10.0
+                    )
+                    if resp.status_code == 200:
+                        # Daemon is alive but SSE is idle — update activity to
+                        # avoid repeated warnings (connection may just be quiet)
+                        self._last_sse_activity = time.time()
+                        logger.debug("Signal: daemon healthy, SSE idle")
+                    else:
+                        logger.warning("Signal: health check failed (%d), forcing reconnect", resp.status_code)
+                        self._force_reconnect()
+                except Exception as e:
+                    logger.warning("Signal: health check error: %s, forcing reconnect", e)
+                    self._force_reconnect()
+
+    def _force_reconnect(self) -> None:
+        """Force SSE reconnection by closing the current response."""
+        if self._sse_response and not self._sse_response.is_stream_consumed:
+            try:
+                asyncio.create_task(self._sse_response.aclose())
+            except Exception:
+                pass
+            self._sse_response = None
+
+    # ------------------------------------------------------------------
+    # Message Handling
+    # ------------------------------------------------------------------
+
+    async def _handle_envelope(self, envelope: dict) -> None:
+        """Process an incoming signal-cli envelope."""
+        # Unwrap nested envelope if present
+        envelope_data = envelope.get("envelope", envelope)
+
+        # Filter syncMessage envelopes (sent transcripts, read receipts, etc.)
+        # signal-cli may set syncMessage to null vs omitting it, so check key existence
+        if "syncMessage" in envelope_data:
+            return
+
+        # Extract sender info
+        sender = (
+            envelope_data.get("sourceNumber")
+            or envelope_data.get("sourceUuid")
+            or envelope_data.get("source")
+        )
+        sender_name = envelope_data.get("sourceName", "")
+        sender_uuid = envelope_data.get("sourceUuid", "")
+
+        if not sender:
+            logger.debug("Signal: ignoring envelope with no sender")
+            return
+
+        # Self-message filtering — prevent reply loops
+        if self._account_normalized and sender == self._account_normalized:
+            return
+
+        # Filter stories
+        if self.ignore_stories and envelope_data.get("storyMessage"):
+            return
+
+        # Get data message — also check editMessage (edited messages contain
+        # their updated dataMessage inside editMessage.dataMessage)
+        data_message = (
+            envelope_data.get("dataMessage")
+            or (envelope_data.get("editMessage") or {}).get("dataMessage")
+        )
+        if not data_message:
+            return
+
+        # Check for group message
+        group_info = data_message.get("groupInfo")
+        group_id = group_info.get("groupId") if group_info else None
+        is_group = bool(group_id)
+
+        # Group message filtering — derived from SIGNAL_GROUP_ALLOWED_USERS:
+        # - No env var set → groups disabled (default safe behavior)
+        # - Env var set with group IDs → only those groups allowed
+        # - Env var set with "*" → all groups allowed
+        # DM auth is fully handled by run.py (_is_user_authorized)
+        if is_group:
+            if not self.group_allow_from:
+                logger.debug("Signal: ignoring group message (no SIGNAL_GROUP_ALLOWED_USERS)")
+                return
+            if "*" not in self.group_allow_from and group_id not in self.group_allow_from:
+                logger.debug("Signal: group %s not in allowlist", group_id[:8] if group_id else "?")
+                return
+
+        # Build chat info
+        chat_id = sender if not is_group else f"group:{group_id}"
+        chat_type = "group" if is_group else "dm"
+
+        # Extract text and render mentions
+        text = data_message.get("message", "")
+        mentions = data_message.get("mentions", [])
+        if text and mentions:
+            text = _render_mentions(text, mentions)
+
+        # Process attachments
+        attachments_data = data_message.get("attachments", [])
+        media_urls = []
+        media_types = []
+
+        if attachments_data and not getattr(self, "ignore_attachments", False):
+            for att in attachments_data:
+                att_id = att.get("id")
+                att_size = att.get("size", 0)
+                if not att_id:
+                    continue
+                if att_size > SIGNAL_MAX_ATTACHMENT_SIZE:
+                    logger.warning("Signal: attachment too large (%d bytes), skipping", att_size)
+                    continue
+                try:
+                    cached_path, ext = await self._fetch_attachment(att_id)
+                    if cached_path:
+                        # Use contentType from Signal if available, else map from extension
+                        content_type = att.get("contentType") or _ext_to_mime(ext)
+                        media_urls.append(cached_path)
+                        media_types.append(content_type)
+                except Exception:
+                    logger.exception("Signal: failed to fetch attachment %s", att_id)
+
+        # Build session source
+        source = self.build_source(
+            chat_id=chat_id,
+            chat_name=group_info.get("groupName") if group_info else sender_name,
+            chat_type=chat_type,
+            user_id=sender,
+            user_name=sender_name or sender,
+            user_id_alt=sender_uuid if sender_uuid else None,
+            chat_id_alt=group_id if is_group else None,
+        )
+
+        # Determine message type from media
+        msg_type = MessageType.TEXT
+        if media_types:
+            if any(mt.startswith("audio/") for mt in media_types):
+                msg_type = MessageType.VOICE
+            elif any(mt.startswith("image/") for mt in media_types):
+                msg_type = MessageType.IMAGE
+
+        # Parse timestamp from envelope data (milliseconds since epoch)
+        ts_ms = envelope_data.get("timestamp", 0)
+        if ts_ms:
+            try:
+                timestamp = datetime.fromtimestamp(ts_ms / 1000, tz=timezone.utc)
+            except (ValueError, OSError):
+                timestamp = datetime.now(tz=timezone.utc)
+        else:
+            timestamp = datetime.now(tz=timezone.utc)
+
+        # Build and dispatch event
+        event = MessageEvent(
+            source=source,
+            text=text or "",
+            message_type=msg_type,
+            media_urls=media_urls,
+            media_types=media_types,
+            timestamp=timestamp,
+        )
+
+        logger.debug("Signal: message from %s in %s: %s",
+                      _redact_phone(sender), chat_id[:20], (text or "")[:50])
+
+        await self.handle_message(event)
+
+    # ------------------------------------------------------------------
+    # Attachment Handling
+    # ------------------------------------------------------------------
+
+    async def _fetch_attachment(self, attachment_id: str) -> tuple:
+        """Fetch an attachment via JSON-RPC and cache it. Returns (path, ext)."""
+        result = await self._rpc("getAttachment", {
+            "account": self.account,
+            "attachmentId": attachment_id,
+        })
+
+        if not result:
+            return None, ""
+
+        # Result is base64-encoded file content
+        raw_data = base64.b64decode(result)
+        ext = _guess_extension(raw_data)
+
+        if _is_image_ext(ext):
+            path = cache_image_from_bytes(raw_data, ext)
+        elif _is_audio_ext(ext):
+            path = cache_audio_from_bytes(raw_data, ext)
+        else:
+            path = cache_document_from_bytes(raw_data, ext)
+
+        return path, ext
+
+    # ------------------------------------------------------------------
+    # JSON-RPC Communication
+    # ------------------------------------------------------------------
+
+    async def _rpc(self, method: str, params: dict, rpc_id: str = None) -> Any:
+        """Send a JSON-RPC 2.0 request to signal-cli daemon."""
+        if not self.client:
+            logger.warning("Signal: RPC called but client not connected")
+            return None
+
+        if rpc_id is None:
+            rpc_id = f"{method}_{int(time.time() * 1000)}"
+
+        payload = {
+            "jsonrpc": "2.0",
+            "method": method,
+            "params": params,
+            "id": rpc_id,
+        }
+
+        try:
+            resp = await self.client.post(
+                f"{self.http_url}/api/v1/rpc",
+                json=payload,
+                timeout=30.0,
+            )
+            resp.raise_for_status()
+            data = resp.json()
+
+            if "error" in data:
+                logger.warning("Signal RPC error (%s): %s", method, data["error"])
+                return None
+
+            return data.get("result")
+
+        except Exception as e:
+            logger.warning("Signal RPC %s failed: %s", method, e)
+            return None
+
+    # ------------------------------------------------------------------
+    # Sending
+    # ------------------------------------------------------------------
+
+    async def send(
+        self,
+        chat_id: str,
+        content: str,
+        reply_to: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None,
+    ) -> SendResult:
+        """Send a text message."""
+        await self._stop_typing_indicator(chat_id)
+
+        params: Dict[str, Any] = {
+            "account": self.account,
+            "message": content,
+        }
+
+        if chat_id.startswith("group:"):
+            params["groupId"] = chat_id[6:]
+        else:
+            params["recipient"] = [chat_id]
+
+        result = await self._rpc("send", params)
+
+        if result is not None:
+            return SendResult(success=True)
+        return SendResult(success=False, error="RPC send failed")
+
+    async def send_typing(self, chat_id: str, metadata=None) -> None:
+        """Send a typing indicator."""
+        params: Dict[str, Any] = {
+            "account": self.account,
+        }
+
+        if chat_id.startswith("group:"):
+            params["groupId"] = chat_id[6:]
+        else:
+            params["recipient"] = [chat_id]
+
+        await self._rpc("sendTyping", params, rpc_id="typing")
+
+    async def send_image(
+        self,
+        chat_id: str,
+        image_url: str,
+        caption: Optional[str] = None,
+        **kwargs,
+    ) -> SendResult:
+        """Send an image. Supports http(s):// and file:// URLs."""
+        await self._stop_typing_indicator(chat_id)
+
+        # Resolve image to local path
+        if image_url.startswith("file://"):
+            file_path = unquote(image_url[7:])
+        else:
+            # Download remote image to cache
+            try:
+                file_path = await cache_image_from_url(image_url)
+            except Exception as e:
+                logger.warning("Signal: failed to download image: %s", e)
+                return SendResult(success=False, error=str(e))
+
+        if not file_path or not Path(file_path).exists():
+            return SendResult(success=False, error="Image file not found")
+
+        # Validate size
+        file_size = Path(file_path).stat().st_size
+        if file_size > SIGNAL_MAX_ATTACHMENT_SIZE:
+            return SendResult(success=False, error=f"Image too large ({file_size} bytes)")
+
+        params: Dict[str, Any] = {
+            "account": self.account,
+            "message": caption or "",
+            "attachments": [file_path],
+        }
+
+        if chat_id.startswith("group:"):
+            params["groupId"] = chat_id[6:]
+        else:
+            params["recipient"] = [chat_id]
+
+        result = await self._rpc("send", params)
+        if result is not None:
+            return SendResult(success=True)
+        return SendResult(success=False, error="RPC send with attachment failed")
+
+    async def send_document(
+        self,
+        chat_id: str,
+        file_path: str,
+        caption: Optional[str] = None,
+        filename: Optional[str] = None,
+        **kwargs,
+    ) -> SendResult:
+        """Send a document/file attachment."""
+        await self._stop_typing_indicator(chat_id)
+
+        if not Path(file_path).exists():
+            return SendResult(success=False, error="File not found")
+
+        params: Dict[str, Any] = {
+            "account": self.account,
+            "message": caption or "",
+            "attachments": [file_path],
+        }
+
+        if chat_id.startswith("group:"):
+            params["groupId"] = chat_id[6:]
+        else:
+            params["recipient"] = [chat_id]
+
+        result = await self._rpc("send", params)
+        if result is not None:
+            return SendResult(success=True)
+        return SendResult(success=False, error="RPC send document failed")
+
+    # ------------------------------------------------------------------
+    # Typing Indicators
+    # ------------------------------------------------------------------
+
+    async def _start_typing_indicator(self, chat_id: str) -> None:
+        """Start a typing indicator loop for a chat."""
+        if chat_id in self._typing_tasks:
+            return  # Already running
+
+        async def _typing_loop():
+            try:
+                while True:
+                    await self.send_typing(chat_id)
+                    await asyncio.sleep(TYPING_INTERVAL)
+            except asyncio.CancelledError:
+                pass
+
+        self._typing_tasks[chat_id] = asyncio.create_task(_typing_loop())
+
+    async def _stop_typing_indicator(self, chat_id: str) -> None:
+        """Stop a typing indicator loop for a chat."""
+        task = self._typing_tasks.pop(chat_id, None)
+        if task:
+            task.cancel()
+            try:
+                await task
+            except asyncio.CancelledError:
+                pass
+
+    # ------------------------------------------------------------------
+    # Chat Info
+    # ------------------------------------------------------------------
+
+    async def get_chat_info(self, chat_id: str) -> Dict[str, Any]:
+        """Get information about a chat/contact."""
+        if chat_id.startswith("group:"):
+            return {
+                "name": chat_id,
+                "type": "group",
+                "chat_id": chat_id,
+            }
+
+        # Try to resolve contact name
+        result = await self._rpc("getContact", {
+            "account": self.account,
+            "contactAddress": chat_id,
+        })
+
+        name = chat_id
+        if result and isinstance(result, dict):
+            name = result.get("name") or result.get("profileName") or chat_id
+
+        return {
+            "name": name,
+            "type": "dm",
+            "chat_id": chat_id,
+        }
--- a/gateway/platforms/slack.py
+++ b/gateway/platforms/slack.py
@@ -10,6 +10,7 @@ Uses slack-bolt (Python) with Socket Mode for:

 import asyncio
 import os
+import re
 from typing import Dict, List, Optional, Any

 try:
@@ -33,6 +34,8 @@ from gateway.platforms.base import (
    MessageEvent,
    MessageType,
    SendResult,
+    SUPPORTED_DOCUMENT_TYPES,
+    cache_document_from_bytes,
    cache_image_from_url,
    cache_audio_from_url,
 )
@@ -96,6 +99,13 @@ class SlackAdapter(BasePlatformAdapter):
            async def handle_message_event(event, say):
                await self._handle_slack_message(event)

+            # Acknowledge app_mention events to prevent Bolt 404 errors.
+            # The "message" handler above already processes @mentions in
+            # channels, so this is intentionally a no-op to avoid duplicates.
+            @self._app.event("app_mention")
+            async def handle_app_mention(event, say):
+                pass
+
            # Register slash command handler
            @self._app.command("/hermes")
            async def handle_hermes_command(ack, command):
@@ -175,7 +185,7 @@ class SlackAdapter(BasePlatformAdapter):
        except Exception as e:
            return SendResult(success=False, error=str(e))

-    async def send_typing(self, chat_id: str) -> None:
+    async def send_typing(self, chat_id: str, metadata=None) -> None:
        """Slack doesn't have a direct typing indicator API for bots."""
        pass

@@ -266,6 +276,65 @@ class SlackAdapter(BasePlatformAdapter):
        except Exception as e:
            return SendResult(success=False, error=str(e))

+    async def send_video(
+        self,
+        chat_id: str,
+        video_path: str,
+        caption: Optional[str] = None,
+        reply_to: Optional[str] = None,
+    ) -> SendResult:
+        """Send a video file to Slack."""
+        if not self._app:
+            return SendResult(success=False, error="Not connected")
+
+        if not os.path.exists(video_path):
+            return SendResult(success=False, error=f"Video file not found: {video_path}")
+
+        try:
+            result = await self._app.client.files_upload_v2(
+                channel=chat_id,
+                file=video_path,
+                filename=os.path.basename(video_path),
+                initial_comment=caption or "",
+                thread_ts=reply_to,
+            )
+            return SendResult(success=True, raw_response=result)
+
+        except Exception as e:
+            print(f"[{self.name}] Failed to send video: {e}")
+            return await super().send_video(chat_id, video_path, caption, reply_to)
+
+    async def send_document(
+        self,
+        chat_id: str,
+        file_path: str,
+        caption: Optional[str] = None,
+        file_name: Optional[str] = None,
+        reply_to: Optional[str] = None,
+    ) -> SendResult:
+        """Send a document/file attachment to Slack."""
+        if not self._app:
+            return SendResult(success=False, error="Not connected")
+
+        if not os.path.exists(file_path):
+            return SendResult(success=False, error=f"File not found: {file_path}")
+
+        display_name = file_name or os.path.basename(file_path)
+
+        try:
+            result = await self._app.client.files_upload_v2(
+                channel=chat_id,
+                file=file_path,
+                filename=display_name,
+                initial_comment=caption or "",
+                thread_ts=reply_to,
+            )
+            return SendResult(success=True, raw_response=result)
+
+        except Exception as e:
+            print(f"[{self.name}] Failed to send document: {e}")
+            return await super().send_document(chat_id, file_path, caption, file_name, reply_to)
+
    async def get_chat_info(self, chat_id: str) -> Dict[str, Any]:
        """Get information about a Slack channel."""
        if not self._app:
@@ -347,6 +416,58 @@ class SlackAdapter(BasePlatformAdapter):
                    msg_type = MessageType.VOICE
                except Exception as e:
                    print(f"[Slack] Failed to cache audio: {e}", flush=True)
+            elif url:
+                # Try to handle as a document attachment
+                try:
+                    original_filename = f.get("name", "")
+                    ext = ""
+                    if original_filename:
+                        _, ext = os.path.splitext(original_filename)
+                        ext = ext.lower()
+
+                    # Fallback: reverse-lookup from MIME type
+                    if not ext and mimetype:
+                        mime_to_ext = {v: k for k, v in SUPPORTED_DOCUMENT_TYPES.items()}
+                        ext = mime_to_ext.get(mimetype, "")
+
+                    if ext not in SUPPORTED_DOCUMENT_TYPES:
+                        continue  # Skip unsupported file types silently
+
+                    # Check file size (Slack limit: 20 MB for bots)
+                    file_size = f.get("size", 0)
+                    MAX_DOC_BYTES = 20 * 1024 * 1024
+                    if not file_size or file_size > MAX_DOC_BYTES:
+                        print(f"[Slack] Document too large or unknown size: {file_size}", flush=True)
+                        continue
+
+                    # Download and cache
+                    raw_bytes = await self._download_slack_file_bytes(url)
+                    cached_path = cache_document_from_bytes(
+                        raw_bytes, original_filename or f"document{ext}"
+                    )
+                    doc_mime = SUPPORTED_DOCUMENT_TYPES[ext]
+                    media_urls.append(cached_path)
+                    media_types.append(doc_mime)
+                    msg_type = MessageType.DOCUMENT
+                    print(f"[Slack] Cached user document: {cached_path}", flush=True)
+
+                    # Inject text content for .txt/.md files (capped at 100 KB)
+                    MAX_TEXT_INJECT_BYTES = 100 * 1024
+                    if ext in (".md", ".txt") and len(raw_bytes) <= MAX_TEXT_INJECT_BYTES:
+                        try:
+                            text_content = raw_bytes.decode("utf-8")
+                            display_name = original_filename or f"document{ext}"
+                            display_name = re.sub(r'[^\w.\- ]', '_', display_name)
+                            injection = f"[Content of {display_name}]:\n{text_content}"
+                            if text:
+                                text = f"{injection}\n\n{text}"
+                            else:
+                                text = injection
+                        except UnicodeDecodeError:
+                            pass  # Binary content, skip injection
+
+                except Exception as e:
+                    print(f"[Slack] Failed to cache document: {e}", flush=True)

        # Build source
        source = self.build_source(
@@ -427,3 +548,16 @@ class SlackAdapter(BasePlatformAdapter):
        else:
            from gateway.platforms.base import cache_image_from_bytes
            return cache_image_from_bytes(response.content, ext)
+
+    async def _download_slack_file_bytes(self, url: str) -> bytes:
+        """Download a Slack file and return raw bytes."""
+        import httpx
+
+        bot_token = self.config.token
+        async with httpx.AsyncClient(timeout=30.0, follow_redirects=True) as client:
+            response = await client.get(
+                url,
+                headers={"Authorization": f"Bearer {bot_token}"},
+            )
+            response.raise_for_status()
+        return response.content
--- a/gateway/platforms/telegram.py
+++ b/gateway/platforms/telegram.py
@@ -86,6 +86,9 @@ def _strip_mdv2(text: str) -> str:
    cleaned = re.sub(r'\\([_*\[\]()~`>#\+\-=|{}.!\\])', r'\1', text)
    # Remove MarkdownV2 bold markers that format_message converted from **bold**
    cleaned = re.sub(r'\*([^*]+)\*', r'\1', cleaned)
+    # Remove MarkdownV2 italic markers that format_message converted from *italic*
+    # Use word boundary (\b) to avoid breaking snake_case like my_variable_name
+    cleaned = re.sub(r'(?<!\w)_([^_]+)_(?!\w)', r'\1', cleaned)
    return cleaned


@@ -132,6 +135,10 @@ class TelegramAdapter(BasePlatformAdapter):
                filters.COMMAND,
                self._handle_command
            ))
+            self._app.add_handler(TelegramMessageHandler(
+                filters.LOCATION | getattr(filters, "VENUE", filters.LOCATION),
+                self._handle_location_message
+            ))
            self._app.add_handler(TelegramMessageHandler(
                filters.PHOTO | filters.VIDEO | filters.AUDIO | filters.VOICE | filters.Document.ALL | filters.Sticker.ALL,
                self._handle_media_message
@@ -282,6 +289,7 @@ class TelegramAdapter(BasePlatformAdapter):
        audio_path: str,
        caption: Optional[str] = None,
        reply_to: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None,
    ) -> SendResult:
        """Send audio as a native Telegram voice message or audio file."""
        if not self._bot:
@@ -295,19 +303,23 @@ class TelegramAdapter(BasePlatformAdapter):
            with open(audio_path, "rb") as audio_file:
                # .ogg files -> send as voice (round playable bubble)
                if audio_path.endswith(".ogg") or audio_path.endswith(".opus"):
+                    _voice_thread = metadata.get("thread_id") if metadata else None
                    msg = await self._bot.send_voice(
                        chat_id=int(chat_id),
                        voice=audio_file,
                        caption=caption[:1024] if caption else None,
                        reply_to_message_id=int(reply_to) if reply_to else None,
+                        message_thread_id=int(_voice_thread) if _voice_thread else None,
                    )
                else:
                    # .mp3 and others -> send as audio file
+                    _audio_thread = metadata.get("thread_id") if metadata else None
                    msg = await self._bot.send_audio(
                        chat_id=int(chat_id),
                        audio=audio_file,
                        caption=caption[:1024] if caption else None,
                        reply_to_message_id=int(reply_to) if reply_to else None,
+                        message_thread_id=int(_audio_thread) if _audio_thread else None,
                    )
            return SendResult(success=True, message_id=str(msg.message_id))
        except Exception as e:
@@ -348,6 +360,7 @@ class TelegramAdapter(BasePlatformAdapter):
        image_url: str,
        caption: Optional[str] = None,
        reply_to: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None,
    ) -> SendResult:
        """Send an image natively as a Telegram photo.
        
@@ -359,11 +372,13 @@ class TelegramAdapter(BasePlatformAdapter):
        
        try:
            # Telegram can send photos directly from URLs (up to ~5MB)
+            _photo_thread = metadata.get("thread_id") if metadata else None
            msg = await self._bot.send_photo(
                chat_id=int(chat_id),
                photo=image_url,
                caption=caption[:1024] if caption else None,  # Telegram caption limit
                reply_to_message_id=int(reply_to) if reply_to else None,
+                message_thread_id=int(_photo_thread) if _photo_thread else None,
            )
            return SendResult(success=True, message_id=str(msg.message_id))
        except Exception as e:
@@ -394,17 +409,20 @@ class TelegramAdapter(BasePlatformAdapter):
        animation_url: str,
        caption: Optional[str] = None,
        reply_to: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None,
    ) -> SendResult:
        """Send an animated GIF natively as a Telegram animation (auto-plays inline)."""
        if not self._bot:
            return SendResult(success=False, error="Not connected")
        
        try:
+            _anim_thread = metadata.get("thread_id") if metadata else None
            msg = await self._bot.send_animation(
                chat_id=int(chat_id),
                animation=animation_url,
                caption=caption[:1024] if caption else None,
                reply_to_message_id=int(reply_to) if reply_to else None,
+                message_thread_id=int(_anim_thread) if _anim_thread else None,
            )
            return SendResult(success=True, message_id=str(msg.message_id))
        except Exception as e:
@@ -412,13 +430,15 @@ class TelegramAdapter(BasePlatformAdapter):
            # Fallback: try as a regular photo
            return await self.send_image(chat_id, animation_url, caption, reply_to)

-    async def send_typing(self, chat_id: str) -> None:
+    async def send_typing(self, chat_id: str, metadata: Optional[Dict[str, Any]] = None) -> None:
        """Send typing indicator."""
        if self._bot:
            try:
+                _typing_thread = metadata.get("thread_id") if metadata else None
                await self._bot.send_chat_action(
                    chat_id=int(chat_id),
-                    action="typing"
+                    action="typing",
+                    message_thread_id=int(_typing_thread) if _typing_thread else None,
                )
            except Exception:
                pass  # Ignore typing indicator failures
@@ -546,6 +566,41 @@ class TelegramAdapter(BasePlatformAdapter):
        event = self._build_message_event(update.message, MessageType.COMMAND)
        await self.handle_message(event)
    
+    async def _handle_location_message(self, update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
+        """Handle incoming location/venue pin messages."""
+        if not update.message:
+            return
+
+        msg = update.message
+        venue = getattr(msg, "venue", None)
+        location = getattr(venue, "location", None) if venue else getattr(msg, "location", None)
+
+        if not location:
+            return
+
+        lat = getattr(location, "latitude", None)
+        lon = getattr(location, "longitude", None)
+        if lat is None or lon is None:
+            return
+
+        # Build a text message with coordinates and context
+        parts = ["[The user shared a location pin.]"]
+        if venue:
+            title = getattr(venue, "title", None)
+            address = getattr(venue, "address", None)
+            if title:
+                parts.append(f"Venue: {title}")
+            if address:
+                parts.append(f"Address: {address}")
+        parts.append(f"latitude: {lat}")
+        parts.append(f"longitude: {lon}")
+        parts.append(f"Map: https://www.google.com/maps/search/?api=1&query={lat},{lon}")
+        parts.append("Ask what they'd like to find nearby (restaurants, cafes, etc.) and any preferences.")
+
+        event = self._build_message_event(msg, MessageType.LOCATION)
+        event.text = "\n".join(parts)
+        await self.handle_message(event)
+
    async def _handle_media_message(self, update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
        """Handle incoming media messages, downloading images to local cache."""
        if not update.message:
--- a/gateway/platforms/whatsapp.py
+++ b/gateway/platforms/whatsapp.py
@@ -493,7 +493,7 @@ class WhatsAppAdapter(BasePlatformAdapter):
            file_name or os.path.basename(file_path),
        )

-    async def send_typing(self, chat_id: str) -> None:
+    async def send_typing(self, chat_id: str, metadata=None) -> None:
        """Send typing indicator via bridge."""
        if not self._running:
            return
--- a/gateway/run.py
+++ b/gateway/run.py
@@ -48,7 +48,7 @@ _config_path = _hermes_home / 'config.yaml'
 if _config_path.exists():
    try:
        import yaml as _yaml
-        with open(_config_path) as _f:
+        with open(_config_path, encoding="utf-8") as _f:
            _cfg = _yaml.safe_load(_f) or {}
        # Top-level simple values (fallback only — don't override .env)
        for _key, _val in _cfg.items():
@@ -75,11 +75,16 @@ if _config_path.exists():
                "container_memory": "TERMINAL_CONTAINER_MEMORY",
                "container_disk": "TERMINAL_CONTAINER_DISK",
                "container_persistent": "TERMINAL_CONTAINER_PERSISTENT",
+                "docker_volumes": "TERMINAL_DOCKER_VOLUMES",
                "sandbox_dir": "TERMINAL_SANDBOX_DIR",
            }
            for _cfg_key, _env_var in _terminal_env_map.items():
                if _cfg_key in _terminal_cfg:
-                    os.environ[_env_var] = str(_terminal_cfg[_cfg_key])
+                    _val = _terminal_cfg[_cfg_key]
+                    if isinstance(_val, list):
+                        os.environ[_env_var] = json.dumps(_val)
+                    else:
+                        os.environ[_env_var] = str(_val)
        _compression_cfg = _cfg.get("compression", {})
        if _compression_cfg and isinstance(_compression_cfg, dict):
            _compression_env_map = {
@@ -118,6 +123,12 @@ if _config_path.exists():
        _tz_cfg = _cfg.get("timezone", "")
        if _tz_cfg and isinstance(_tz_cfg, str) and "HERMES_TIMEZONE" not in os.environ:
            os.environ["HERMES_TIMEZONE"] = _tz_cfg.strip()
+        # Security settings
+        _security_cfg = _cfg.get("security", {})
+        if isinstance(_security_cfg, dict):
+            _redact = _security_cfg.get("redact_secrets")
+            if _redact is not None:
+                os.environ["HERMES_REDACT_SECRETS"] = str(_redact).lower()
    except Exception:
        pass  # Non-fatal; gateway can still run with .env values

@@ -194,6 +205,7 @@ class GatewayRunner:
        self._ephemeral_system_prompt = self._load_ephemeral_system_prompt()
        self._reasoning_config = self._load_reasoning_config()
        self._provider_routing = self._load_provider_routing()
+        self._fallback_model = self._load_fallback_model()

        # Wire process registry into session store for reset protection
        from tools.process_registry import process_registry
@@ -304,7 +316,7 @@ class GatewayRunner:
                import yaml as _y
                cfg_path = _hermes_home / "config.yaml"
                if cfg_path.exists():
-                    with open(cfg_path) as _f:
+                    with open(cfg_path, encoding="utf-8") as _f:
                        cfg = _y.safe_load(_f) or {}
                    file_path = cfg.get("prefill_messages_file", "")
            except Exception:
@@ -342,7 +354,7 @@ class GatewayRunner:
            import yaml as _y
            cfg_path = _hermes_home / "config.yaml"
            if cfg_path.exists():
-                with open(cfg_path) as _f:
+                with open(cfg_path, encoding="utf-8") as _f:
                    cfg = _y.safe_load(_f) or {}
                return (cfg.get("agent", {}).get("system_prompt", "") or "").strip()
        except Exception:
@@ -363,7 +375,7 @@ class GatewayRunner:
                import yaml as _y
                cfg_path = _hermes_home / "config.yaml"
                if cfg_path.exists():
-                    with open(cfg_path) as _f:
+                    with open(cfg_path, encoding="utf-8") as _f:
                        cfg = _y.safe_load(_f) or {}
                    effort = str(cfg.get("agent", {}).get("reasoning_effort", "") or "").strip()
            except Exception:
@@ -379,6 +391,41 @@ class GatewayRunner:
        logger.warning("Unknown reasoning_effort '%s', using default (medium)", effort)
        return None

+    @staticmethod
+    def _load_background_notifications_mode() -> str:
+        """Load background process notification mode from config or env var.
+
+        Modes:
+          - ``all``    — push running-output updates *and* the final message (default)
+          - ``result`` — only the final completion message (regardless of exit code)
+          - ``error``  — only the final message when exit code is non-zero
+          - ``off``    — no watcher messages at all
+        """
+        mode = os.getenv("HERMES_BACKGROUND_NOTIFICATIONS", "")
+        if not mode:
+            try:
+                import yaml as _y
+                cfg_path = _hermes_home / "config.yaml"
+                if cfg_path.exists():
+                    with open(cfg_path, encoding="utf-8") as _f:
+                        cfg = _y.safe_load(_f) or {}
+                    raw = cfg.get("display", {}).get("background_process_notifications")
+                    if raw is False:
+                        mode = "off"
+                    elif raw not in (None, ""):
+                        mode = str(raw)
+            except Exception:
+                pass
+        mode = (mode or "all").strip().lower()
+        valid = {"all", "result", "error", "off"}
+        if mode not in valid:
+            logger.warning(
+                "Unknown background_process_notifications '%s', defaulting to 'all'",
+                mode,
+            )
+            return "all"
+        return mode
+
    @staticmethod
    def _load_provider_routing() -> dict:
        """Load OpenRouter provider routing preferences from config.yaml."""
@@ -386,13 +433,33 @@ class GatewayRunner:
            import yaml as _y
            cfg_path = _hermes_home / "config.yaml"
            if cfg_path.exists():
-                with open(cfg_path) as _f:
+                with open(cfg_path, encoding="utf-8") as _f:
                    cfg = _y.safe_load(_f) or {}
                return cfg.get("provider_routing", {}) or {}
        except Exception:
            pass
        return {}

+    @staticmethod
+    def _load_fallback_model() -> dict | None:
+        """Load fallback model config from config.yaml.
+
+        Returns a dict with 'provider' and 'model' keys, or None if
+        not configured / both fields empty.
+        """
+        try:
+            import yaml as _y
+            cfg_path = _hermes_home / "config.yaml"
+            if cfg_path.exists():
+                with open(cfg_path, encoding="utf-8") as _f:
+                    cfg = _y.safe_load(_f) or {}
+                fb = cfg.get("fallback_model", {}) or {}
+                if fb.get("provider") and fb.get("model"):
+                    return fb
+        except Exception:
+            pass
+        return None
+
    async def start(self) -> bool:
        """
        Start the gateway and all configured platform adapters.
@@ -591,6 +658,13 @@ class GatewayRunner:
                return None
            return SlackAdapter(config)

+        elif platform == Platform.SIGNAL:
+            from gateway.platforms.signal import SignalAdapter, check_signal_requirements
+            if not check_signal_requirements():
+                logger.warning("Signal: SIGNAL_HTTP_URL or SIGNAL_ACCOUNT not configured")
+                return None
+            return SignalAdapter(config)
+
        elif platform == Platform.HOMEASSISTANT:
            from gateway.platforms.homeassistant import HomeAssistantAdapter, check_ha_requirements
            if not check_ha_requirements():
@@ -626,12 +700,14 @@ class GatewayRunner:
            Platform.DISCORD: "DISCORD_ALLOWED_USERS",
            Platform.WHATSAPP: "WHATSAPP_ALLOWED_USERS",
            Platform.SLACK: "SLACK_ALLOWED_USERS",
+            Platform.SIGNAL: "SIGNAL_ALLOWED_USERS",
        }
        platform_allow_all_map = {
            Platform.TELEGRAM: "TELEGRAM_ALLOW_ALL_USERS",
            Platform.DISCORD: "DISCORD_ALLOW_ALL_USERS",
            Platform.WHATSAPP: "WHATSAPP_ALLOW_ALL_USERS",
            Platform.SLACK: "SLACK_ALLOW_ALL_USERS",
+            Platform.SIGNAL: "SIGNAL_ALLOW_ALL_USERS",
        }

        # Per-platform allow-all flag (e.g., DISCORD_ALLOW_ALL_USERS=true)
@@ -730,7 +806,7 @@ class GatewayRunner:
        _known_commands = {"new", "reset", "help", "status", "stop", "model",
                          "personality", "retry", "undo", "sethome", "set-home",
                          "compress", "usage", "insights", "reload-mcp", "reload_mcp",
-                          "update", "title", "resume", "provider"}
+                          "update", "title", "resume", "provider", "rollback"}
        if command and command in _known_commands:
            await self.hooks.emit(f"command:{command}", {
                "platform": source.platform.value if source.platform else "",
@@ -789,6 +865,9 @@ class GatewayRunner:

        if command == "resume":
            return await self._handle_resume_command(event)
+
+        if command == "rollback":
+            return await self._handle_rollback_command(event)
        
        # Skill slash commands: /skill-name loads the skill and sends to agent
        if command:
@@ -870,159 +949,187 @@ class GatewayRunner:
        # every new message rehydrates an oversized transcript, causing
        # repeated truncation/context failures.  Detect this early and
        # compress proactively — before the agent even starts.  (#628)
+        #
+        # Thresholds are derived from the SAME compression config the
+        # agent uses (compression.threshold × model context length) so
+        # CLI and messaging platforms behave identically.
        # -----------------------------------------------------------------
        if history and len(history) >= 4:
-            from agent.model_metadata import estimate_messages_tokens_rough
+            from agent.model_metadata import (
+                estimate_messages_tokens_rough,
+                get_model_context_length,
+            )

-            # Read thresholds from config.yaml → session_hygiene section
-            _hygiene_cfg = {}
+            # Read model + compression config from config.yaml — same
+            # source of truth the agent itself uses.
+            _hyg_model = "anthropic/claude-sonnet-4.6"
+            _hyg_threshold_pct = 0.85
+            _hyg_compression_enabled = True
            try:
                _hyg_cfg_path = _hermes_home / "config.yaml"
                if _hyg_cfg_path.exists():
                    import yaml as _hyg_yaml
-                    with open(_hyg_cfg_path) as _hyg_f:
+                    with open(_hyg_cfg_path, encoding="utf-8") as _hyg_f:
                        _hyg_data = _hyg_yaml.safe_load(_hyg_f) or {}
-                    _hygiene_cfg = _hyg_data.get("session_hygiene", {})
-                    if not isinstance(_hygiene_cfg, dict):
-                        _hygiene_cfg = {}
+
+                    # Resolve model name (same logic as run_sync)
+                    _model_cfg = _hyg_data.get("model", {})
+                    if isinstance(_model_cfg, str):
+                        _hyg_model = _model_cfg
+                    elif isinstance(_model_cfg, dict):
+                        _hyg_model = _model_cfg.get("default", _hyg_model)
+
+                    # Read compression settings
+                    _comp_cfg = _hyg_data.get("compression", {})
+                    if isinstance(_comp_cfg, dict):
+                        _hyg_threshold_pct = float(
+                            _comp_cfg.get("threshold", _hyg_threshold_pct)
+                        )
+                        _hyg_compression_enabled = str(
+                            _comp_cfg.get("enabled", True)
+                        ).lower() in ("true", "1", "yes")
            except Exception:
                pass

-            _compress_token_threshold = int(
-                _hygiene_cfg.get("auto_compress_tokens", 100_000)
-            )
-            _compress_msg_threshold = int(
-                _hygiene_cfg.get("auto_compress_messages", 200)
-            )
-            _warn_token_threshold = int(
-                _hygiene_cfg.get("warn_tokens", 200_000)
+            # Also check env overrides (same as run_agent.py)
+            _hyg_threshold_pct = float(
+                os.getenv("CONTEXT_COMPRESSION_THRESHOLD", str(_hyg_threshold_pct))
            )
+            if os.getenv("CONTEXT_COMPRESSION_ENABLED", "").lower() in ("false", "0", "no"):
+                _hyg_compression_enabled = False

-            _msg_count = len(history)
-            _approx_tokens = estimate_messages_tokens_rough(history)
-
-            _needs_compress = (
-                _approx_tokens >= _compress_token_threshold
-                or _msg_count >= _compress_msg_threshold
-            )
-
-            if _needs_compress:
-                logger.info(
-                    "Session hygiene: %s messages, ~%s tokens — auto-compressing "
-                    "(thresholds: %s msgs / %s tokens)",
-                    _msg_count, f"{_approx_tokens:,}",
-                    _compress_msg_threshold, f"{_compress_token_threshold:,}",
+            if _hyg_compression_enabled:
+                _hyg_context_length = get_model_context_length(_hyg_model)
+                _compress_token_threshold = int(
+                    _hyg_context_length * _hyg_threshold_pct
                )
+                # Warn if still huge after compression (95% of context)
+                _warn_token_threshold = int(_hyg_context_length * 0.95)
+
+                _msg_count = len(history)
+                _approx_tokens = estimate_messages_tokens_rough(history)
+
+                _needs_compress = _approx_tokens >= _compress_token_threshold
+
+                if _needs_compress:
+                    logger.info(
+                        "Session hygiene: %s messages, ~%s tokens — auto-compressing "
+                        "(threshold: %s%% of %s = %s tokens)",
+                        _msg_count, f"{_approx_tokens:,}",
+                        int(_hyg_threshold_pct * 100),
+                        f"{_hyg_context_length:,}",
+                        f"{_compress_token_threshold:,}",
+                    )
+
+                    _hyg_adapter = self.adapters.get(source.platform)
+                    if _hyg_adapter:
+                        try:
+                            await _hyg_adapter.send(
+                                source.chat_id,
+                                f"🗜️ Session is large ({_msg_count} messages, "
+                                f"~{_approx_tokens:,} tokens). Auto-compressing..."
+                            )
+                        except Exception:
+                            pass

-                _hyg_adapter = self.adapters.get(source.platform)
-                if _hyg_adapter:
                    try:
-                        await _hyg_adapter.send(
-                            source.chat_id,
-                            f"🗜️ Session is large ({_msg_count} messages, "
-                            f"~{_approx_tokens:,} tokens). Auto-compressing..."
-                        )
-                    except Exception:
-                        pass
+                        from run_agent import AIAgent

-                try:
-                    from run_agent import AIAgent
+                        _hyg_runtime = _resolve_runtime_agent_kwargs()
+                        if _hyg_runtime.get("api_key"):
+                            _hyg_msgs = [
+                                {"role": m.get("role"), "content": m.get("content")}
+                                for m in history
+                                if m.get("role") in ("user", "assistant")
+                                and m.get("content")
+                            ]

-                    _hyg_runtime = _resolve_runtime_agent_kwargs()
-                    if _hyg_runtime.get("api_key"):
-                        _hyg_msgs = [
-                            {"role": m.get("role"), "content": m.get("content")}
-                            for m in history
-                            if m.get("role") in ("user", "assistant")
-                            and m.get("content")
-                        ]
-
-                        if len(_hyg_msgs) >= 4:
-                            _hyg_agent = AIAgent(
-                                **_hyg_runtime,
-                                max_iterations=4,
-                                quiet_mode=True,
-                                enabled_toolsets=["memory"],
-                                session_id=session_entry.session_id,
-                            )
-
-                            loop = asyncio.get_event_loop()
-                            _compressed, _ = await loop.run_in_executor(
-                                None,
-                                lambda: _hyg_agent._compress_context(
-                                    _hyg_msgs, "",
-                                    approx_tokens=_approx_tokens,
-                                ),
-                            )
-
-                            self.session_store.rewrite_transcript(
-                                session_entry.session_id, _compressed
-                            )
-                            history = _compressed
-                            _new_count = len(_compressed)
-                            _new_tokens = estimate_messages_tokens_rough(
-                                _compressed
-                            )
-
-                            logger.info(
-                                "Session hygiene: compressed %s → %s msgs, "
-                                "~%s → ~%s tokens",
-                                _msg_count, _new_count,
-                                f"{_approx_tokens:,}", f"{_new_tokens:,}",
-                            )
-
-                            if _hyg_adapter:
-                                try:
-                                    await _hyg_adapter.send(
-                                        source.chat_id,
-                                        f"🗜️ Compressed: {_msg_count} → "
-                                        f"{_new_count} messages, "
-                                        f"~{_approx_tokens:,} → "
-                                        f"~{_new_tokens:,} tokens"
-                                    )
-                                except Exception:
-                                    pass
-
-                            # Still too large after compression — warn user
-                            if _new_tokens >= _warn_token_threshold:
-                                logger.warning(
-                                    "Session hygiene: still ~%s tokens after "
-                                    "compression — suggesting /reset",
-                                    f"{_new_tokens:,}",
+                            if len(_hyg_msgs) >= 4:
+                                _hyg_agent = AIAgent(
+                                    **_hyg_runtime,
+                                    max_iterations=4,
+                                    quiet_mode=True,
+                                    enabled_toolsets=["memory"],
+                                    session_id=session_entry.session_id,
                                )
+
+                                loop = asyncio.get_event_loop()
+                                _compressed, _ = await loop.run_in_executor(
+                                    None,
+                                    lambda: _hyg_agent._compress_context(
+                                        _hyg_msgs, "",
+                                        approx_tokens=_approx_tokens,
+                                    ),
+                                )
+
+                                self.session_store.rewrite_transcript(
+                                    session_entry.session_id, _compressed
+                                )
+                                history = _compressed
+                                _new_count = len(_compressed)
+                                _new_tokens = estimate_messages_tokens_rough(
+                                    _compressed
+                                )
+
+                                logger.info(
+                                    "Session hygiene: compressed %s → %s msgs, "
+                                    "~%s → ~%s tokens",
+                                    _msg_count, _new_count,
+                                    f"{_approx_tokens:,}", f"{_new_tokens:,}",
+                                )
+
                                if _hyg_adapter:
                                    try:
                                        await _hyg_adapter.send(
                                            source.chat_id,
-                                            "⚠️ Session is still very large "
-                                            "after compression "
-                                            f"(~{_new_tokens:,} tokens). "
-                                            "Consider using /reset to start "
-                                            "fresh if you experience issues."
+                                            f"🗜️ Compressed: {_msg_count} → "
+                                            f"{_new_count} messages, "
+                                            f"~{_approx_tokens:,} → "
+                                            f"~{_new_tokens:,} tokens"
                                        )
                                    except Exception:
                                        pass

-                except Exception as e:
-                    logger.warning(
-                        "Session hygiene auto-compress failed: %s", e
-                    )
-                    # Compression failed and session is dangerously large
-                    if _approx_tokens >= _warn_token_threshold:
-                        _hyg_adapter = self.adapters.get(source.platform)
-                        if _hyg_adapter:
-                            try:
-                                await _hyg_adapter.send(
-                                    source.chat_id,
-                                    f"⚠️ Session is very large "
-                                    f"({_msg_count} messages, "
-                                    f"~{_approx_tokens:,} tokens) and "
-                                    "auto-compression failed. Consider "
-                                    "using /compress or /reset to avoid "
-                                    "issues."
-                                )
-                            except Exception:
-                                pass
+                                # Still too large after compression — warn user
+                                if _new_tokens >= _warn_token_threshold:
+                                    logger.warning(
+                                        "Session hygiene: still ~%s tokens after "
+                                        "compression — suggesting /reset",
+                                        f"{_new_tokens:,}",
+                                    )
+                                    if _hyg_adapter:
+                                        try:
+                                            await _hyg_adapter.send(
+                                                source.chat_id,
+                                                "⚠️ Session is still very large "
+                                                "after compression "
+                                                f"(~{_new_tokens:,} tokens). "
+                                                "Consider using /reset to start "
+                                                "fresh if you experience issues."
+                                            )
+                                        except Exception:
+                                            pass
+
+                    except Exception as e:
+                        logger.warning(
+                            "Session hygiene auto-compress failed: %s", e
+                        )
+                        # Compression failed and session is dangerously large
+                        if _approx_tokens >= _warn_token_threshold:
+                            _hyg_adapter = self.adapters.get(source.platform)
+                            if _hyg_adapter:
+                                try:
+                                    await _hyg_adapter.send(
+                                        source.chat_id,
+                                        f"⚠️ Session is very large "
+                                        f"({_msg_count} messages, "
+                                        f"~{_approx_tokens:,} tokens) and "
+                                        "auto-compression failed. Consider "
+                                        "using /compress or /reset to avoid "
+                                        "issues."
+                                    )
+                                except Exception:
+                                    pass

        # First-message onboarding -- only on the very first interaction ever
        if not history and not self.session_store.has_any_sessions():
@@ -1215,6 +1322,11 @@ class GatewayRunner:
                        {"role": "assistant", "content": response, "timestamp": ts}
                    )
            else:
+                # The agent already persisted these messages to SQLite via
+                # _flush_messages_to_session_db(), so skip the DB write here
+                # to prevent the duplicate-write bug (#860).  We still write
+                # to JSONL for backward compatibility and as a backup.
+                agent_persisted = self._session_db is not None
                for msg in new_messages:
                    # Skip system messages (they're rebuilt each run)
                    if msg.get("role") == "system":
@@ -1222,7 +1334,8 @@ class GatewayRunner:
                    # Add timestamp to each message for debugging
                    entry = {**msg, "timestamp": ts}
                    self.session_store.append_to_transcript(
-                        session_entry.session_id, entry
+                        session_entry.session_id, entry,
+                        skip_db=agent_persisted,
                    )
            
            # Update session
@@ -1331,6 +1444,7 @@ class GatewayRunner:
            "`/resume [name]` — Resume a previously-named session",
            "`/usage` — Show token usage for this session",
            "`/insights [days]` — Show usage insights and analytics",
+            "`/rollback [number]` — List or restore filesystem checkpoints",
            "`/reload-mcp` — Reload MCP servers from config",
            "`/update` — Update Hermes Agent to the latest version",
            "`/help` — Show this message",
@@ -1365,7 +1479,7 @@ class GatewayRunner:
        current_provider = "openrouter"
        try:
            if config_path.exists():
-                with open(config_path) as f:
+                with open(config_path, encoding="utf-8") as f:
                    cfg = yaml.safe_load(f) or {}
                model_cfg = cfg.get("model", {})
                if isinstance(model_cfg, str):
@@ -1385,6 +1499,11 @@ class GatewayRunner:
            except Exception:
                current_provider = "openrouter"

+        # Detect custom endpoint: provider resolved to openrouter but a custom
+        # base URL is configured — the user set up a custom endpoint.
+        if current_provider == "openrouter" and os.getenv("OPENAI_BASE_URL", "").strip():
+            current_provider = "custom"
+
        if not args:
            provider_label = _PROVIDER_LABELS.get(current_provider, current_provider)
            lines = [
@@ -1451,14 +1570,14 @@ class GatewayRunner:
            try:
                user_config = {}
                if config_path.exists():
-                    with open(config_path) as f:
+                    with open(config_path, encoding="utf-8") as f:
                        user_config = yaml.safe_load(f) or {}
                if "model" not in user_config or not isinstance(user_config["model"], dict):
                    user_config["model"] = {}
                user_config["model"]["default"] = new_model
                if provider_changed:
                    user_config["model"]["provider"] = target_provider
-                with open(config_path, 'w') as f:
+                with open(config_path, 'w', encoding="utf-8") as f:
                    yaml.dump(user_config, f, default_flow_style=False, sort_keys=False)
            except Exception as e:
                return f"⚠️ Failed to save model change: {e}"
@@ -1495,7 +1614,7 @@ class GatewayRunner:
        config_path = _hermes_home / 'config.yaml'
        try:
            if config_path.exists():
-                with open(config_path) as f:
+                with open(config_path, encoding="utf-8") as f:
                    cfg = yaml.safe_load(f) or {}
                model_cfg = cfg.get("model", {})
                if isinstance(model_cfg, dict):
@@ -1511,6 +1630,10 @@ class GatewayRunner:
            except Exception:
                current_provider = "openrouter"

+        # Detect custom endpoint
+        if current_provider == "openrouter" and os.getenv("OPENAI_BASE_URL", "").strip():
+            current_provider = "custom"
+
        current_label = _PROVIDER_LABELS.get(current_provider, current_provider)

        lines = [
@@ -1540,7 +1663,7 @@ class GatewayRunner:

        try:
            if config_path.exists():
-                with open(config_path, 'r') as f:
+                with open(config_path, 'r', encoding="utf-8") as f:
                    config = yaml.safe_load(f) or {}
                personalities = config.get("agent", {}).get("personalities", {})
            else:
@@ -1569,7 +1692,7 @@ class GatewayRunner:
                if "agent" not in config or not isinstance(config.get("agent"), dict):
                    config["agent"] = {}
                config["agent"]["system_prompt"] = new_prompt
-                with open(config_path, 'w') as f:
+                with open(config_path, 'w', encoding="utf-8") as f:
                    yaml.dump(config, f, default_flow_style=False, sort_keys=False)
            except Exception as e:
                return f"⚠️ Failed to save personality change: {e}"
@@ -1653,10 +1776,10 @@ class GatewayRunner:
            config_path = _hermes_home / 'config.yaml'
            user_config = {}
            if config_path.exists():
-                with open(config_path) as f:
+                with open(config_path, encoding="utf-8") as f:
                    user_config = yaml.safe_load(f) or {}
            user_config[env_key] = chat_id
-            with open(config_path, 'w') as f:
+            with open(config_path, 'w', encoding="utf-8") as f:
                yaml.dump(user_config, f, default_flow_style=False)
            # Also set in the current environment so it takes effect immediately
            os.environ[env_key] = str(chat_id)
@@ -1668,6 +1791,65 @@ class GatewayRunner:
            f"Cron jobs and cross-platform messages will be delivered here."
        )
    
+    async def _handle_rollback_command(self, event: MessageEvent) -> str:
+        """Handle /rollback command — list or restore filesystem checkpoints."""
+        from tools.checkpoint_manager import CheckpointManager, format_checkpoint_list
+
+        # Read checkpoint config from config.yaml
+        cp_cfg = {}
+        try:
+            import yaml as _y
+            _cfg_path = _hermes_home / "config.yaml"
+            if _cfg_path.exists():
+                with open(_cfg_path, encoding="utf-8") as _f:
+                    _data = _y.safe_load(_f) or {}
+                cp_cfg = _data.get("checkpoints", {})
+                if isinstance(cp_cfg, bool):
+                    cp_cfg = {"enabled": cp_cfg}
+        except Exception:
+            pass
+
+        if not cp_cfg.get("enabled", False):
+            return (
+                "Checkpoints are not enabled.\n"
+                "Enable in config.yaml:\n```\ncheckpoints:\n  enabled: true\n```"
+            )
+
+        mgr = CheckpointManager(
+            enabled=True,
+            max_snapshots=cp_cfg.get("max_snapshots", 50),
+        )
+
+        cwd = os.getenv("MESSAGING_CWD", str(Path.home()))
+        arg = event.get_command_args().strip()
+
+        if not arg:
+            checkpoints = mgr.list_checkpoints(cwd)
+            return format_checkpoint_list(checkpoints, cwd)
+
+        # Restore by number or hash
+        checkpoints = mgr.list_checkpoints(cwd)
+        if not checkpoints:
+            return f"No checkpoints found for {cwd}"
+
+        target_hash = None
+        try:
+            idx = int(arg) - 1
+            if 0 <= idx < len(checkpoints):
+                target_hash = checkpoints[idx]["hash"]
+            else:
+                return f"Invalid checkpoint number. Use 1-{len(checkpoints)}."
+        except ValueError:
+            target_hash = arg
+
+        result = mgr.restore(cwd, target_hash)
+        if result["success"]:
+            return (
+                f"✅ Restored to checkpoint {result['restored_to']}: {result['reason']}\n"
+                f"A pre-rollback snapshot was saved automatically."
+            )
+        return f"❌ {result['error']}"
+
    async def _handle_compress_command(self, event: MessageEvent) -> str:
        """Handle /compress command -- manually compress conversation context."""
        source = event.source
@@ -2229,6 +2411,12 @@ class GatewayRunner:

        Runs as an asyncio task. Stays silent when nothing changed.
        Auto-removes when the process exits or is killed.
+
+        Notification mode (from ``display.background_process_notifications``):
+          - ``all``    — running-output updates + final message
+          - ``result`` — final completion message only
+          - ``error``  — final message only when exit code != 0
+          - ``off``    — no messages at all
        """
        from tools.process_registry import process_registry

@@ -2237,8 +2425,21 @@ class GatewayRunner:
        session_key = watcher.get("session_key", "")
        platform_name = watcher.get("platform", "")
        chat_id = watcher.get("chat_id", "")
+        notify_mode = self._load_background_notifications_mode()

-        logger.debug("Process watcher started: %s (every %ss)", session_id, interval)
+        logger.debug("Process watcher started: %s (every %ss, notify=%s)",
+                      session_id, interval, notify_mode)
+
+        if notify_mode == "off":
+            # Still wait for the process to exit so we can log it, but don't
+            # push any messages to the user.
+            while True:
+                await asyncio.sleep(interval)
+                session = process_registry.get(session_id)
+                if session is None or session.exited:
+                    break
+            logger.debug("Process watcher ended (silent): %s", session_id)
+            return

        last_output_len = 0
        while True:
@@ -2253,27 +2454,31 @@ class GatewayRunner:
            last_output_len = current_output_len

            if session.exited:
-                # Process finished -- deliver final update
-                new_output = session.output_buffer[-1000:] if session.output_buffer else ""
-                message_text = (
-                    f"[Background process {session_id} finished with exit code {session.exit_code}~ "
-                    f"Here's the final output:\n{new_output}]"
+                # Decide whether to notify based on mode
+                should_notify = (
+                    notify_mode in ("all", "result")
+                    or (notify_mode == "error" and session.exit_code not in (0, None))
                )
-                # Try to deliver to the originating platform
-                adapter = None
-                for p, a in self.adapters.items():
-                    if p.value == platform_name:
-                        adapter = a
-                        break
-                if adapter and chat_id:
-                    try:
-                        await adapter.send(chat_id, message_text)
-                    except Exception as e:
-                        logger.error("Watcher delivery error: %s", e)
+                if should_notify:
+                    new_output = session.output_buffer[-1000:] if session.output_buffer else ""
+                    message_text = (
+                        f"[Background process {session_id} finished with exit code {session.exit_code}~ "
+                        f"Here's the final output:\n{new_output}]"
+                    )
+                    adapter = None
+                    for p, a in self.adapters.items():
+                        if p.value == platform_name:
+                            adapter = a
+                            break
+                    if adapter and chat_id:
+                        try:
+                            await adapter.send(chat_id, message_text)
+                        except Exception as e:
+                            logger.error("Watcher delivery error: %s", e)
                break

-            elif has_new_output:
-                # New output available -- deliver status update
+            elif has_new_output and notify_mode == "all":
+                # New output available -- deliver status update (only in "all" mode)
                new_output = session.output_buffer[-500:] if session.output_buffer else ""
                message_text = (
                    f"[Background process {session_id} is still running~ "
@@ -2324,6 +2529,8 @@ class GatewayRunner:
            Platform.DISCORD: "hermes-discord",
            Platform.WHATSAPP: "hermes-whatsapp",
            Platform.SLACK: "hermes-slack",
+            Platform.SIGNAL: "hermes-signal",
+            Platform.HOMEASSISTANT: "hermes-homeassistant",
        }
        
        # Try to load platform_toolsets from config
@@ -2332,7 +2539,7 @@ class GatewayRunner:
            config_path = _hermes_home / 'config.yaml'
            if config_path.exists():
                import yaml
-                with open(config_path, 'r') as f:
+                with open(config_path, 'r', encoding="utf-8") as f:
                    user_config = yaml.safe_load(f) or {}
                platform_toolsets_config = user_config.get("platform_toolsets", {})
        except Exception as e:
@@ -2345,6 +2552,8 @@ class GatewayRunner:
            Platform.DISCORD: "discord",
            Platform.WHATSAPP: "whatsapp",
            Platform.SLACK: "slack",
+            Platform.SIGNAL: "signal",
+            Platform.HOMEASSISTANT: "homeassistant",
        }.get(source.platform, "telegram")
        
        # Use config override if present (list of toolsets), otherwise hardcoded default
@@ -2362,7 +2571,7 @@ class GatewayRunner:
            _tp_cfg_path = _hermes_home / "config.yaml"
            if _tp_cfg_path.exists():
                import yaml as _tp_yaml
-                with open(_tp_cfg_path) as _tp_f:
+                with open(_tp_cfg_path, encoding="utf-8") as _tp_f:
                    _tp_data = _tp_yaml.safe_load(_tp_f) or {}
                _progress_cfg = _tp_data.get("display", {})
        except Exception:
@@ -2453,6 +2662,8 @@ class GatewayRunner:
        
        # Background task to send progress messages
        # Accumulates tool lines into a single message that gets edited
+        _progress_metadata = {"thread_id": source.thread_id} if source.thread_id else None
+
        async def send_progress_messages():
            if not progress_queue:
                return
@@ -2482,15 +2693,15 @@ class GatewayRunner:
                            # Platform doesn't support editing — stop trying,
                            # send just this new line as a separate message
                            can_edit = False
-                            await adapter.send(chat_id=source.chat_id, content=msg)
+                            await adapter.send(chat_id=source.chat_id, content=msg, metadata=_progress_metadata)
                    else:
                        if can_edit:
                            # First tool: send all accumulated text as new message
                            full_text = "\n".join(progress_lines)
-                            result = await adapter.send(chat_id=source.chat_id, content=full_text)
+                            result = await adapter.send(chat_id=source.chat_id, content=full_text, metadata=_progress_metadata)
                        else:
                            # Editing unsupported: send just this line
-                            result = await adapter.send(chat_id=source.chat_id, content=msg)
+                            result = await adapter.send(chat_id=source.chat_id, content=msg, metadata=_progress_metadata)
                        if result.success and result.message_id:
                            progress_msg_id = result.message_id

@@ -2580,7 +2791,7 @@ class GatewayRunner:
                import yaml as _y
                _cfg_path = _hermes_home / "config.yaml"
                if _cfg_path.exists():
-                    with open(_cfg_path) as _f:
+                    with open(_cfg_path, encoding="utf-8") as _f:
                        _cfg = _y.safe_load(_f) or {}
                    _model_cfg = _cfg.get("model", {})
                    if isinstance(_model_cfg, str):
@@ -2623,6 +2834,7 @@ class GatewayRunner:
                platform=platform_key,
                honcho_session_key=session_key,
                session_db=self._session_db,
+                fallback_model=self._fallback_model,
            )
            
            # Store agent reference for interrupt support
@@ -3061,7 +3273,7 @@ def main():
    config = None
    if args.config:
        import json
-        with open(args.config) as f:
+        with open(args.config, encoding="utf-8") as f:
            data = json.load(f)
            config = GatewayConfig.from_dict(data)
    
--- a/gateway/session.py
+++ b/gateway/session.py
@@ -45,6 +45,8 @@ class SessionSource:
    user_name: Optional[str] = None
    thread_id: Optional[str] = None  # For forum topics, Discord threads, etc.
    chat_topic: Optional[str] = None  # Channel topic/description (Discord, Slack)
+    user_id_alt: Optional[str] = None  # Signal UUID (alternative to phone number)
+    chat_id_alt: Optional[str] = None  # Signal group internal ID
    
    @property
    def description(self) -> str:
@@ -68,7 +70,7 @@ class SessionSource:
        return ", ".join(parts)
    
    def to_dict(self) -> Dict[str, Any]:
-        return {
+        d = {
            "platform": self.platform.value,
            "chat_id": self.chat_id,
            "chat_name": self.chat_name,
@@ -78,6 +80,11 @@ class SessionSource:
            "thread_id": self.thread_id,
            "chat_topic": self.chat_topic,
        }
+        if self.user_id_alt:
+            d["user_id_alt"] = self.user_id_alt
+        if self.chat_id_alt:
+            d["chat_id_alt"] = self.chat_id_alt
+        return d
    
    @classmethod
    def from_dict(cls, data: Dict[str, Any]) -> "SessionSource":
@@ -90,6 +97,8 @@ class SessionSource:
            user_name=data.get("user_name"),
            thread_id=data.get("thread_id"),
            chat_topic=data.get("chat_topic"),
+            user_id_alt=data.get("user_id_alt"),
+            chat_id_alt=data.get("chat_id_alt"),
        )
    
    @classmethod
@@ -263,8 +272,8 @@ class SessionEntry:
        if data.get("platform"):
            try:
                platform = Platform(data["platform"])
-            except ValueError:
-                pass
+            except ValueError as e:
+                logger.debug("Unknown platform value %r: %s", data["platform"], e)
        
        return cls(
            session_key=data["session_key"],
@@ -333,7 +342,7 @@ class SessionStore:
        
        if sessions_file.exists():
            try:
-                with open(sessions_file, "r") as f:
+                with open(sessions_file, "r", encoding="utf-8") as f:
                    data = json.load(f)
                    for key, entry_data in data.items():
                        self._entries[key] = SessionEntry.from_dict(entry_data)
@@ -344,12 +353,26 @@ class SessionStore:
    
    def _save(self) -> None:
        """Save sessions index to disk (kept for session key -> ID mapping)."""
+        import tempfile
        self.sessions_dir.mkdir(parents=True, exist_ok=True)
        sessions_file = self.sessions_dir / "sessions.json"
-        
+
        data = {key: entry.to_dict() for key, entry in self._entries.items()}
-        with open(sessions_file, "w") as f:
-            json.dump(data, f, indent=2)
+        fd, tmp_path = tempfile.mkstemp(
+            dir=str(self.sessions_dir), suffix=".tmp", prefix=".sessions_"
+        )
+        try:
+            with os.fdopen(fd, "w", encoding="utf-8") as f:
+                json.dump(data, f, indent=2)
+                f.flush()
+                os.fsync(f.fileno())
+            os.replace(tmp_path, sessions_file)
+        except BaseException:
+            try:
+                os.unlink(tmp_path)
+            except OSError as e:
+                logger.debug("Could not remove temp file %s: %s", tmp_path, e)
+            raise
    
    def _generate_session_key(self, source: SessionSource) -> str:
        """Generate a session key from a source."""
@@ -654,10 +677,17 @@ class SessionStore:
        """Get the path to a session's legacy transcript file."""
        return self.sessions_dir / f"{session_id}.jsonl"
    
-    def append_to_transcript(self, session_id: str, message: Dict[str, Any]) -> None:
-        """Append a message to a session's transcript (SQLite + legacy JSONL)."""
-        # Write to SQLite
-        if self._db:
+    def append_to_transcript(self, session_id: str, message: Dict[str, Any], skip_db: bool = False) -> None:
+        """Append a message to a session's transcript (SQLite + legacy JSONL).
+
+        Args:
+            skip_db: When True, only write to JSONL and skip the SQLite write.
+                     Used when the agent already persisted messages to SQLite
+                     via its own _flush_messages_to_session_db(), preventing
+                     the duplicate-write bug (#860).
+        """
+        # Write to SQLite (unless the agent already handled it)
+        if self._db and not skip_db:
            try:
                self._db.append_message(
                    session_id=session_id,
@@ -672,7 +702,7 @@ class SessionStore:
        
        # Also write legacy JSONL (keeps existing tooling working during transition)
        transcript_path = self.get_transcript_path(session_id)
-        with open(transcript_path, "a") as f:
+        with open(transcript_path, "a", encoding="utf-8") as f:
            f.write(json.dumps(message, ensure_ascii=False) + "\n")
    
    def rewrite_transcript(self, session_id: str, messages: List[Dict[str, Any]]) -> None:
@@ -699,7 +729,7 @@ class SessionStore:
        
        # JSONL: overwrite the file
        transcript_path = self.get_transcript_path(session_id)
-        with open(transcript_path, "w") as f:
+        with open(transcript_path, "w", encoding="utf-8") as f:
            for msg in messages:
                f.write(json.dumps(msg, ensure_ascii=False) + "\n")

@@ -721,7 +751,7 @@ class SessionStore:
            return []
        
        messages = []
-        with open(transcript_path, "r") as f:
+        with open(transcript_path, "r", encoding="utf-8") as f:
            for line in f:
                line = line.strip()
                if line:
--- a/hermes_cli/auth.py
+++ b/hermes_cli/auth.py
@@ -23,6 +23,7 @@ import stat
 import base64
 import hashlib
 import subprocess
+import threading
 import time
 import uuid
 import webbrowser
@@ -44,6 +45,10 @@ try:
    import fcntl
 except Exception:
    fcntl = None
+try:
+    import msvcrt
+except Exception:
+    msvcrt = None

 # =============================================================================
 # Constants
@@ -103,6 +108,14 @@ PROVIDER_REGISTRY: Dict[str, ProviderConfig] = {
        auth_type="oauth_external",
        inference_base_url=DEFAULT_CODEX_BASE_URL,
    ),
+    "nous-api": ProviderConfig(
+        id="nous-api",
+        name="Nous Portal (API Key)",
+        auth_type="api_key",
+        inference_base_url="https://inference-api.nousresearch.com/v1",
+        api_key_env_vars=("NOUS_API_KEY",),
+        base_url_env_var="NOUS_BASE_URL",
+    ),
    "zai": ProviderConfig(
        id="zai",
        name="Z.AI / GLM",
@@ -299,31 +312,64 @@ def _auth_lock_path() -> Path:
    return _auth_file_path().with_suffix(".lock")


+_auth_lock_holder = threading.local()
+
@contextmanager
 def _auth_store_lock(timeout_seconds: float = AUTH_LOCK_TIMEOUT_SECONDS):
-    """Cross-process advisory lock for auth.json reads+writes."""
+    """Cross-process advisory lock for auth.json reads+writes.  Reentrant."""
+    # Reentrant: if this thread already holds the lock, just yield.
+    if getattr(_auth_lock_holder, "depth", 0) > 0:
+        _auth_lock_holder.depth += 1
+        try:
+            yield
+        finally:
+            _auth_lock_holder.depth -= 1
+        return
+
    lock_path = _auth_lock_path()
    lock_path.parent.mkdir(parents=True, exist_ok=True)

-    with lock_path.open("a+") as lock_file:
-        if fcntl is None:
+    if fcntl is None and msvcrt is None:
+        _auth_lock_holder.depth = 1
+        try:
            yield
-            return
+        finally:
+            _auth_lock_holder.depth = 0
+        return

+    # On Windows, msvcrt.locking needs the file to have content and the
+    # file pointer at position 0.  Ensure the lock file has at least 1 byte.
+    if msvcrt and (not lock_path.exists() or lock_path.stat().st_size == 0):
+        lock_path.write_text(" ", encoding="utf-8")
+
+    with lock_path.open("r+" if msvcrt else "a+") as lock_file:
        deadline = time.time() + max(1.0, timeout_seconds)
        while True:
            try:
-                fcntl.flock(lock_file.fileno(), fcntl.LOCK_EX | fcntl.LOCK_NB)
+                if fcntl:
+                    fcntl.flock(lock_file.fileno(), fcntl.LOCK_EX | fcntl.LOCK_NB)
+                else:
+                    lock_file.seek(0)
+                    msvcrt.locking(lock_file.fileno(), msvcrt.LK_NBLCK, 1)
                break
-            except BlockingIOError:
+            except (BlockingIOError, OSError, PermissionError):
                if time.time() >= deadline:
                    raise TimeoutError("Timed out waiting for auth store lock")
                time.sleep(0.05)

+        _auth_lock_holder.depth = 1
        try:
            yield
        finally:
-            fcntl.flock(lock_file.fileno(), fcntl.LOCK_UN)
+            _auth_lock_holder.depth = 0
+            if fcntl:
+                fcntl.flock(lock_file.fileno(), fcntl.LOCK_UN)
+            elif msvcrt:
+                try:
+                    lock_file.seek(0)
+                    msvcrt.locking(lock_file.fileno(), msvcrt.LK_UNLCK, 1)
+                except (OSError, IOError):
+                    pass


 def _load_auth_store(auth_file: Optional[Path] = None) -> Dict[str, Any]:
@@ -475,6 +521,7 @@ def resolve_provider(

    # Normalize provider aliases
    _PROVIDER_ALIASES = {
+        "nous_api": "nous-api", "nousapi": "nous-api", "nous-portal-api": "nous-api",
        "glm": "zai", "z-ai": "zai", "z.ai": "zai", "zhipu": "zai",
        "kimi": "kimi-coding", "moonshot": "kimi-coding",
        "minimax-china": "minimax-cn", "minimax_cn": "minimax-cn",
@@ -1056,6 +1103,19 @@ def fetch_nous_models(
                continue
            model_ids.append(mid)

+    # Sort: prefer opus > pro > haiku/flash > sonnet (sonnet is cheap/fast,
+    # users who want the best model should see opus first).
+    def _model_priority(mid: str) -> tuple:
+        low = mid.lower()
+        if "opus" in low:
+            return (0, mid)
+        if "pro" in low and "sonnet" not in low:
+            return (1, mid)
+        if "sonnet" in low:
+            return (3, mid)
+        return (2, mid)
+
+    model_ids.sort(key=_model_priority)
    return list(dict.fromkeys(model_ids))


@@ -1624,11 +1684,11 @@ def _save_model_choice(model_id: str) -> None:
    from hermes_cli.config import save_config, load_config, save_env_value

    config = load_config()
-    # Handle both string and dict model formats
+    # Always use dict format so provider/base_url can be stored alongside
    if isinstance(config.get("model"), dict):
        config["model"]["default"] = model_id
    else:
-        config["model"] = model_id
+        config["model"] = {"default": model_id}
    save_config(config)
    save_env_value("LLM_MODEL", model_id)

--- a/hermes_cli/banner.py
+++ b/hermes_cli/banner.py
@@ -36,6 +36,28 @@ def cprint(text: str):
    _pt_print(_PT_ANSI(text))


+# =========================================================================
+# Skin-aware color helpers
+# =========================================================================
+
+def _skin_color(key: str, fallback: str) -> str:
+    """Get a color from the active skin, or return fallback."""
+    try:
+        from hermes_cli.skin_engine import get_active_skin
+        return get_active_skin().get_color(key, fallback)
+    except Exception:
+        return fallback
+
+
+def _skin_branding(key: str, fallback: str) -> str:
+    """Get a branding string from the active skin, or return fallback."""
+    try:
+        from hermes_cli.skin_engine import get_active_skin
+        return get_active_skin().get_branding(key, fallback)
+    except Exception:
+        return fallback
+
+
 # =========================================================================
 # ASCII Art & Branding
 # =========================================================================
@@ -217,18 +239,24 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
    layout_table.add_column("left", justify="center")
    layout_table.add_column("right", justify="left")

+    # Resolve skin colors once for the entire banner
+    accent = _skin_color("banner_accent", "#FFBF00")
+    dim = _skin_color("banner_dim", "#B8860B")
+    text = _skin_color("banner_text", "#FFF8DC")
+    session_color = _skin_color("session_border", "#8B8682")
+
    left_lines = ["", HERMES_CADUCEUS, ""]
    model_short = model.split("/")[-1] if "/" in model else model
    if len(model_short) > 28:
        model_short = model_short[:25] + "..."
-    ctx_str = f" [dim #B8860B]·[/] [dim #B8860B]{_format_context_length(context_length)} context[/]" if context_length else ""
-    left_lines.append(f"[#FFBF00]{model_short}[/]{ctx_str} [dim #B8860B]·[/] [dim #B8860B]Nous Research[/]")
-    left_lines.append(f"[dim #B8860B]{cwd}[/]")
+    ctx_str = f" [dim {dim}]·[/] [dim {dim}]{_format_context_length(context_length)} context[/]" if context_length else ""
+    left_lines.append(f"[{accent}]{model_short}[/]{ctx_str} [dim {dim}]·[/] [dim {dim}]Nous Research[/]")
+    left_lines.append(f"[dim {dim}]{cwd}[/]")
    if session_id:
-        left_lines.append(f"[dim #8B8682]Session: {session_id}[/]")
+        left_lines.append(f"[dim {session_color}]Session: {session_id}[/]")
    left_content = "\n".join(left_lines)

-    right_lines = ["[bold #FFBF00]Available Tools[/]"]
+    right_lines = [f"[bold {accent}]Available Tools[/]"]
    toolsets_dict: Dict[str, list] = {}

    for tool in tools:
@@ -256,7 +284,7 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
            if name in disabled_tools:
                colored_names.append(f"[red]{name}[/]")
            else:
-                colored_names.append(f"[#FFF8DC]{name}[/]")
+                colored_names.append(f"[{text}]{name}[/]")

        tools_str = ", ".join(colored_names)
        if len(", ".join(sorted(tool_names))) > 45:
@@ -275,7 +303,7 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
                elif name in disabled_tools:
                    colored_names.append(f"[red]{name}[/]")
                else:
-                    colored_names.append(f"[#FFF8DC]{name}[/]")
+                    colored_names.append(f"[{text}]{name}[/]")
            tools_str = ", ".join(colored_names)

        right_lines.append(f"[dim #B8860B]{toolset}:[/] {tools_str}")
@@ -306,7 +334,7 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
                )

    right_lines.append("")
-    right_lines.append("[bold #FFBF00]Available Skills[/]")
+    right_lines.append(f"[bold {accent}]Available Skills[/]")
    skills_by_category = get_available_skills()
    total_skills = sum(len(s) for s in skills_by_category.values())

@@ -320,9 +348,9 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
                skills_str = ", ".join(skill_names)
            if len(skills_str) > 50:
                skills_str = skills_str[:47] + "..."
-            right_lines.append(f"[dim #B8860B]{category}:[/] [#FFF8DC]{skills_str}[/]")
+            right_lines.append(f"[dim {dim}]{category}:[/] [{text}]{skills_str}[/]")
    else:
-        right_lines.append("[dim #B8860B]No skills installed[/]")
+        right_lines.append(f"[dim {dim}]No skills installed[/]")

    right_lines.append("")
    mcp_connected = sum(1 for s in mcp_status if s["connected"]) if mcp_status else 0
@@ -330,7 +358,7 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
    if mcp_connected:
        summary_parts.append(f"{mcp_connected} MCP servers")
    summary_parts.append("/help for commands")
-    right_lines.append(f"[dim #B8860B]{' · '.join(summary_parts)}[/]")
+    right_lines.append(f"[dim {dim}]{' · '.join(summary_parts)}[/]")

    # Update check — show if behind origin/main
    try:
@@ -347,10 +375,13 @@ def build_welcome_banner(console: Console, model: str, cwd: str,
    right_content = "\n".join(right_lines)
    layout_table.add_row(left_content, right_content)

+    agent_name = _skin_branding("agent_name", "Hermes Agent")
+    title_color = _skin_color("banner_title", "#FFD700")
+    border_color = _skin_color("banner_border", "#CD7F32")
    outer_panel = Panel(
        layout_table,
-        title=f"[bold #FFD700]Hermes Agent {VERSION}[/]",
-        border_style="#CD7F32",
+        title=f"[bold {title_color}]{agent_name} {VERSION}[/]",
+        border_style=border_color,
        padding=(0, 2),
    )

--- a/hermes_cli/clipboard.py
+++ b/hermes_cli/clipboard.py
@@ -292,9 +292,12 @@ def _convert_to_png(path: Path) -> bool:
            ["convert", str(tmp), "png:" + str(path)],
            capture_output=True, timeout=5,
        )
-        tmp.unlink(missing_ok=True)
        if r.returncode == 0 and path.exists() and path.stat().st_size > 0:
+            tmp.unlink(missing_ok=True)
            return True
+        else:
+            # Convert failed — restore the original file
+            tmp.rename(path)
    except FileNotFoundError:
        logger.debug("ImageMagick not installed — cannot convert BMP to PNG")
        if tmp.exists() and not path.exists():
--- a/hermes_cli/codex_models.py
+++ b/hermes_cli/codex_models.py
@@ -47,7 +47,7 @@ def _fetch_models_from_api(access_token: str) -> List[str]:
        if item.get("supported_in_api") is False:
            continue
        visibility = item.get("visibility", "")
-        if isinstance(visibility, str) and visibility.strip().lower() == "hide":
+        if isinstance(visibility, str) and visibility.strip().lower() == "hidden":
            continue
        priority = item.get("priority")
        rank = int(priority) if isinstance(priority, (int, float)) else 10_000
--- a/hermes_cli/commands.py
+++ b/hermes_cli/commands.py
@@ -39,6 +39,8 @@ COMMANDS = {
    "/insights": "Show usage insights and analytics (last 30 days)",
    "/paste": "Check clipboard for an image and attach it",
    "/reload-mcp": "Reload MCP servers from config.yaml",
+    "/rollback": "List or restore filesystem checkpoints (usage: /rollback [number])",
+    "/skin": "Show or change the display skin/theme",
    "/quit": "Exit the CLI (also: /exit, /q)",
 }

--- a/hermes_cli/config.py
+++ b/hermes_cli/config.py
@@ -14,8 +14,9 @@ This module provides:

 import os
 import platform
-import sys
+import stat
 import subprocess
+import sys
 from pathlib import Path
 from typing import Dict, Any, Optional, List, Tuple

@@ -62,7 +63,9 @@ def ensure_hermes_home():
 DEFAULT_CONFIG = {
    "model": "anthropic/claude-opus-4.6",
    "toolsets": ["hermes-cli"],
-    "max_turns": 100,
+    "agent": {
+        "max_turns": 90,
+    },
    
    "terminal": {
        "backend": "local",
@@ -77,10 +80,23 @@ DEFAULT_CONFIG = {
        "container_memory": 5120,       # MB (default 5GB)
        "container_disk": 51200,        # MB (default 50GB)
        "container_persistent": True,   # Persist filesystem across sessions
+        # Docker volume mounts — share host directories with the container.
+        # Each entry is "host_path:container_path" (standard Docker -v syntax).
+        # Example: ["/home/user/projects:/workspace/projects", "/data:/data"]
+        "docker_volumes": [],
    },
    
    "browser": {
        "inactivity_timeout": 120,
+        "record_sessions": False,  # Auto-record browser sessions as WebM videos
+    },
+    
+    # Filesystem checkpoints — automatic snapshots before destructive file ops.
+    # When enabled, the agent takes a snapshot of the working directory once per
+    # conversation turn (on first write_file/patch call).  Use /rollback to restore.
+    "checkpoints": {
+        "enabled": False,
+        "max_snapshots": 50,  # Max checkpoints to keep per directory
    },
    
    "compression": {
@@ -106,8 +122,9 @@ DEFAULT_CONFIG = {
    "display": {
        "compact": False,
        "personality": "kawaii",
-        "resume_display": "full",  # "full" (show previous messages) | "minimal" (one-liner only)
-        "bell_on_complete": False,  # Play terminal bell (\a) when agent finishes a response
+        "resume_display": "full",
+        "bell_on_complete": False,
+        "skin": "default",
    },
    
    # Text-to-speech configuration
@@ -165,7 +182,7 @@ DEFAULT_CONFIG = {
    "command_allowlist": [],

    # Config schema version - bump this when adding new required fields
-    "_config_version": 5,
+    "_config_version": 6,
 }

 # =============================================================================
@@ -190,6 +207,22 @@ REQUIRED_ENV_VARS = {}
 # Optional environment variables that enhance functionality
 OPTIONAL_ENV_VARS = {
    # ── Provider (handled in provider selection, not shown in checklists) ──
+    "NOUS_API_KEY": {
+        "description": "Nous Portal API key (direct API key access to Nous inference)",
+        "prompt": "Nous Portal API key",
+        "url": "https://portal.nousresearch.com",
+        "password": True,
+        "category": "provider",
+        "advanced": True,
+    },
+    "NOUS_BASE_URL": {
+        "description": "Nous Portal base URL override",
+        "prompt": "Nous Portal base URL (leave empty for default)",
+        "url": None,
+        "password": False,
+        "category": "provider",
+        "advanced": True,
+    },
    "OPENROUTER_API_KEY": {
        "description": "OpenRouter API key (for vision, web scraping helpers, and MoA)",
        "prompt": "OpenRouter API key",
@@ -400,14 +433,18 @@ OPTIONAL_ENV_VARS = {
        "category": "messaging",
    },
    "SLACK_BOT_TOKEN": {
-        "description": "Slack bot integration",
+        "description": "Slack bot token (xoxb-). Get from OAuth & Permissions after installing your app. "
+                       "Required scopes: chat:write, app_mentions:read, channels:history, groups:history, "
+                       "im:history, im:read, im:write, users:read, files:write",
        "prompt": "Slack Bot Token (xoxb-...)",
        "url": "https://api.slack.com/apps",
        "password": True,
        "category": "messaging",
    },
    "SLACK_APP_TOKEN": {
-        "description": "Slack Socket Mode connection",
+        "description": "Slack app-level token (xapp-) for Socket Mode. Get from Basic Information → "
+                       "App-Level Tokens. Also ensure Event Subscriptions include: message.im, "
+                       "message.channels, message.groups, app_mention",
        "prompt": "Slack App Token (xapp-...)",
        "url": "https://api.slack.com/apps",
        "password": True,
@@ -438,7 +475,7 @@ OPTIONAL_ENV_VARS = {
        "category": "setting",
    },
    "HERMES_MAX_ITERATIONS": {
-        "description": "Maximum tool-calling iterations per conversation (default: 60)",
+        "description": "Maximum tool-calling iterations per conversation (default: 90)",
        "prompt": "Max iterations",
        "url": None,
        "password": False,
@@ -739,6 +776,23 @@ def _deep_merge(base: dict, override: dict) -> dict:
    return result


+def _normalize_max_turns_config(config: Dict[str, Any]) -> Dict[str, Any]:
+    """Normalize legacy root-level max_turns into agent.max_turns."""
+    config = dict(config)
+    agent_config = dict(config.get("agent") or {})
+
+    if "max_turns" in config and "max_turns" not in agent_config:
+        agent_config["max_turns"] = config["max_turns"]
+
+    if "max_turns" not in agent_config:
+        agent_config["max_turns"] = DEFAULT_CONFIG["agent"]["max_turns"]
+
+    config["agent"] = agent_config
+    config.pop("max_turns", None)
+    return config
+
+
+
 def load_config() -> Dict[str, Any]:
    """Load configuration from ~/.hermes/config.yaml."""
    import copy
@@ -748,23 +802,76 @@ def load_config() -> Dict[str, Any]:
    
    if config_path.exists():
        try:
-            with open(config_path) as f:
+            with open(config_path, encoding="utf-8") as f:
                user_config = yaml.safe_load(f) or {}
-            
+
+            if "max_turns" in user_config:
+                agent_user_config = dict(user_config.get("agent") or {})
+                if agent_user_config.get("max_turns") is None:
+                    agent_user_config["max_turns"] = user_config["max_turns"]
+                user_config["agent"] = agent_user_config
+                user_config.pop("max_turns", None)
+
            config = _deep_merge(config, user_config)
        except Exception as e:
            print(f"Warning: Failed to load config: {e}")
    
-    return config
+    return _normalize_max_turns_config(config)
+
+
+_COMMENTED_SECTIONS = """
+# ── Security ──────────────────────────────────────────────────────────
+# API keys, tokens, and passwords are redacted from tool output by default.
+# Set to false to see full values (useful for debugging auth issues).
+#
+# security:
+#   redact_secrets: false
+
+# ── Fallback Model ────────────────────────────────────────────────────
+# Automatic provider failover when primary is unavailable.
+# Uncomment and configure to enable. Triggers on rate limits (429),
+# overload (529), service errors (503), or connection failures.
+#
+# Supported providers:
+#   openrouter   (OPENROUTER_API_KEY)  — routes to any model
+#   openai-codex (OAuth — hermes login) — OpenAI Codex
+#   nous         (OAuth — hermes login) — Nous Portal
+#   zai          (ZAI_API_KEY)         — Z.AI / GLM
+#   kimi-coding  (KIMI_API_KEY)        — Kimi / Moonshot
+#   minimax      (MINIMAX_API_KEY)     — MiniMax
+#   minimax-cn   (MINIMAX_CN_API_KEY)  — MiniMax (China)
+#
+# For custom OpenAI-compatible endpoints, add base_url and api_key_env.
+#
+# fallback_model:
+#   provider: openrouter
+#   model: anthropic/claude-sonnet-4
+"""


 def save_config(config: Dict[str, Any]):
    """Save configuration to ~/.hermes/config.yaml."""
+    from utils import atomic_yaml_write
+
    ensure_hermes_home()
    config_path = get_config_path()
-    
-    with open(config_path, 'w') as f:
-        yaml.dump(config, f, default_flow_style=False, sort_keys=False)
+    normalized = _normalize_max_turns_config(config)
+
+    # Build optional commented-out sections for features that are off by
+    # default or only relevant when explicitly configured.
+    sections = []
+    sec = normalized.get("security", {})
+    if not sec or sec.get("redact_secrets") is None:
+        sections.append("security")
+    fb = normalized.get("fallback_model", {})
+    if not fb or not (fb.get("provider") and fb.get("model")):
+        sections.append("fallback")
+
+    atomic_yaml_write(
+        config_path,
+        normalized,
+        extra_content=_COMMENTED_SECTIONS if sections else None,
+    )


 def load_env() -> Dict[str, str]:
@@ -818,6 +925,13 @@ def save_env_value(key: str, value: str):
    with open(env_path, 'w', **write_kw) as f:
        f.writelines(lines)

+    # Restrict .env permissions to owner-only (contains API keys)
+    if not _IS_WINDOWS:
+        try:
+            os.chmod(env_path, stat.S_IRUSR | stat.S_IWUSR)
+        except OSError:
+            pass
+

 def get_env_value(key: str) -> Optional[str]:
    """Get a value from ~/.hermes/.env or environment."""
@@ -881,7 +995,7 @@ def show_config():
    print()
    print(color("◆ Model", Colors.CYAN, Colors.BOLD))
    print(f"  Model:        {config.get('model', 'not set')}")
-    print(f"  Max turns:    {config.get('max_turns', 100)}")
+    print(f"  Max turns:    {config.get('agent', {}).get('max_turns', DEFAULT_CONFIG['agent']['max_turns'])}")
    print(f"  Toolsets:     {', '.join(config.get('toolsets', ['all']))}")
    
    # Terminal
@@ -1010,7 +1124,7 @@ def set_config_value(key: str, value: str):
        'FAL_KEY', 'TELEGRAM_BOT_TOKEN', 'DISCORD_BOT_TOKEN',
        'TERMINAL_SSH_HOST', 'TERMINAL_SSH_USER', 'TERMINAL_SSH_KEY',
        'SUDO_PASSWORD', 'SLACK_BOT_TOKEN', 'SLACK_APP_TOKEN',
-        'GITHUB_TOKEN', 'HONCHO_API_KEY', 'NOUS_API_KEY', 'WANDB_API_KEY',
+        'GITHUB_TOKEN', 'HONCHO_API_KEY', 'WANDB_API_KEY',
        'TINKER_API_KEY',
    ]
    
@@ -1026,7 +1140,7 @@ def set_config_value(key: str, value: str):
    user_config = {}
    if config_path.exists():
        try:
-            with open(config_path) as f:
+            with open(config_path, encoding="utf-8") as f:
                user_config = yaml.safe_load(f) or {}
        except Exception:
            user_config = {}
@@ -1054,7 +1168,7 @@ def set_config_value(key: str, value: str):
    
    # Write only user config back (not the full merged defaults)
    ensure_hermes_home()
-    with open(config_path, 'w') as f:
+    with open(config_path, 'w', encoding="utf-8") as f:
        yaml.dump(user_config, f, default_flow_style=False, sort_keys=False)
    
    # Keep .env in sync for keys that terminal_tool reads directly from env vars.
--- a/hermes_cli/gateway.py
+++ b/hermes_cli/gateway.py
@@ -482,14 +482,19 @@ _PLATFORMS = [
        "token_var": "SLACK_BOT_TOKEN",
        "setup_instructions": [
            "1. Go to https://api.slack.com/apps → Create New App → From Scratch",
-            "2. Enable Socket Mode: App Settings → Socket Mode → Enable",
-            "3. Get Bot Token: OAuth & Permissions → Install to Workspace → copy xoxb-... token",
-            "4. Get App Token: Basic Information → App-Level Tokens → Generate",
-            "   Name it anything, add scope: connections:write → copy xapp-... token",
-            "5. Add bot scopes: OAuth & Permissions → Scopes → chat:write, im:history,",
-            "   im:read, im:write, channels:history, channels:read",
-            "6. Reinstall the app to your workspace after adding scopes",
+            "2. Enable Socket Mode: Settings → Socket Mode → Enable",
+            "   Create an App-Level Token with scope: connections:write → copy xapp-... token",
+            "3. Add Bot Token Scopes: Features → OAuth & Permissions → Scopes",
+            "   Required: chat:write, app_mentions:read, channels:history, channels:read,",
+            "   groups:history, im:history, im:read, im:write, users:read, files:write",
+            "4. Subscribe to Events: Features → Event Subscriptions → Enable",
+            "   Required events: message.im, message.channels, app_mention",
+            "   Optional: message.groups (for private channels)",
+            "   ⚠ Without message.channels the bot will ONLY work in DMs!",
+            "5. Install to Workspace: Settings → Install App → copy xoxb-... token",
+            "6. Reinstall the app after any scope or event changes",
            "7. Find your user ID: click your profile → three dots → Copy member ID",
+            "8. Invite the bot to channels: /invite @YourBot",
        ],
        "vars": [
            {"name": "SLACK_BOT_TOKEN", "prompt": "Bot Token (xoxb-...)", "password": True,
@@ -507,6 +512,12 @@ _PLATFORMS = [
        "emoji": "📲",
        "token_var": "WHATSAPP_ENABLED",
    },
+    {
+        "key": "signal",
+        "label": "Signal",
+        "emoji": "📡",
+        "token_var": "SIGNAL_HTTP_URL",
+    },
 ]


@@ -525,6 +536,13 @@ def _platform_status(platform: dict) -> str:
                return "configured + paired"
            return "enabled, not paired"
        return "not configured"
+    if platform.get("key") == "signal":
+        account = get_env_value("SIGNAL_ACCOUNT")
+        if val and account:
+            return "configured"
+        if val or account:
+            return "partially configured"
+        return "not configured"
    if val:
        return "configured"
    return "not configured"
@@ -650,6 +668,121 @@ def _is_service_running() -> bool:
    return len(find_gateway_pids()) > 0


+def _setup_signal():
+    """Interactive setup for Signal messenger."""
+    import shutil
+
+    print()
+    print(color("  ─── 📡 Signal Setup ───", Colors.CYAN))
+
+    existing_url = get_env_value("SIGNAL_HTTP_URL")
+    existing_account = get_env_value("SIGNAL_ACCOUNT")
+    if existing_url and existing_account:
+        print()
+        print_success("Signal is already configured.")
+        if not prompt_yes_no("  Reconfigure Signal?", False):
+            return
+
+    # Check if signal-cli is available
+    print()
+    if shutil.which("signal-cli"):
+        print_success("signal-cli found on PATH.")
+    else:
+        print_warning("signal-cli not found on PATH.")
+        print_info("  Signal requires signal-cli running as an HTTP daemon.")
+        print_info("  Install options:")
+        print_info("    Linux:  sudo apt install signal-cli")
+        print_info("            or download from https://github.com/AsamK/signal-cli")
+        print_info("    macOS:  brew install signal-cli")
+        print_info("    Docker: bbernhard/signal-cli-rest-api")
+        print()
+        print_info("  After installing, link your account and start the daemon:")
+        print_info("    signal-cli link -n \"HermesAgent\"")
+        print_info("    signal-cli --account +YOURNUMBER daemon --http 127.0.0.1:8080")
+        print()
+
+    # HTTP URL
+    print()
+    print_info("  Enter the URL where signal-cli HTTP daemon is running.")
+    default_url = existing_url or "http://127.0.0.1:8080"
+    try:
+        url = input(f"  HTTP URL [{default_url}]: ").strip() or default_url
+    except (EOFError, KeyboardInterrupt):
+        print("\n  Setup cancelled.")
+        return
+
+    # Test connectivity
+    print_info("  Testing connection...")
+    try:
+        import httpx
+        resp = httpx.get(f"{url.rstrip('/')}/api/v1/check", timeout=10.0)
+        if resp.status_code == 200:
+            print_success("  signal-cli daemon is reachable!")
+        else:
+            print_warning(f"  signal-cli responded with status {resp.status_code}.")
+            if not prompt_yes_no("  Continue anyway?", False):
+                return
+    except Exception as e:
+        print_warning(f"  Could not reach signal-cli at {url}: {e}")
+        if not prompt_yes_no("  Save this URL anyway? (you can start signal-cli later)", True):
+            return
+
+    save_env_value("SIGNAL_HTTP_URL", url)
+
+    # Account phone number
+    print()
+    print_info("  Enter your Signal account phone number in E.164 format.")
+    print_info("  Example: +15551234567")
+    default_account = existing_account or ""
+    try:
+        account = input(f"  Account number{f' [{default_account}]' if default_account else ''}: ").strip()
+        if not account:
+            account = default_account
+    except (EOFError, KeyboardInterrupt):
+        print("\n  Setup cancelled.")
+        return
+
+    if not account:
+        print_error("  Account number is required.")
+        return
+
+    save_env_value("SIGNAL_ACCOUNT", account)
+
+    # Allowed users
+    print()
+    print_info("  The gateway DENIES all users by default for security.")
+    print_info("  Enter phone numbers or UUIDs of allowed users (comma-separated).")
+    existing_allowed = get_env_value("SIGNAL_ALLOWED_USERS") or ""
+    default_allowed = existing_allowed or account
+    try:
+        allowed = input(f"  Allowed users [{default_allowed}]: ").strip() or default_allowed
+    except (EOFError, KeyboardInterrupt):
+        print("\n  Setup cancelled.")
+        return
+
+    save_env_value("SIGNAL_ALLOWED_USERS", allowed)
+
+    # Group messaging
+    print()
+    if prompt_yes_no("  Enable group messaging? (disabled by default for security)", False):
+        print()
+        print_info("  Enter group IDs to allow, or * for all groups.")
+        existing_groups = get_env_value("SIGNAL_GROUP_ALLOWED_USERS") or ""
+        try:
+            groups = input(f"  Group IDs [{existing_groups or '*'}]: ").strip() or existing_groups or "*"
+        except (EOFError, KeyboardInterrupt):
+            print("\n  Setup cancelled.")
+            return
+        save_env_value("SIGNAL_GROUP_ALLOWED_USERS", groups)
+
+    print()
+    print_success("Signal configured!")
+    print_info(f"  URL: {url}")
+    print_info(f"  Account: {account}")
+    print_info(f"  DM auth: via SIGNAL_ALLOWED_USERS + DM pairing")
+    print_info(f"  Groups: {'enabled' if get_env_value('SIGNAL_GROUP_ALLOWED_USERS') else 'disabled'}")
+
+
 def gateway_setup():
    """Interactive setup for messaging platforms + gateway service."""

@@ -702,6 +835,8 @@ def gateway_setup():

        if platform["key"] == "whatsapp":
            _setup_whatsapp()
+        elif platform["key"] == "signal":
+            _setup_signal()
        else:
            _setup_standard_platform(platform)

--- a/hermes_cli/main.py
+++ b/hermes_cli/main.py
@@ -489,6 +489,7 @@ def cmd_chat(args):
        "query": args.query,
        "resume": getattr(args, "resume", None),
        "worktree": getattr(args, "worktree", False),
+        "checkpoints": getattr(args, "checkpoints", False),
    }
    # Filter out None values
    kwargs = {k: v for k, v in kwargs.items() if v is not None}
@@ -761,9 +762,39 @@ def cmd_model(args):
        ("kimi-coding", "Kimi / Moonshot (Moonshot AI direct API)"),
        ("minimax", "MiniMax (global direct API)"),
        ("minimax-cn", "MiniMax China (domestic direct API)"),
-        ("custom", "Custom endpoint (self-hosted / VLLM / etc.)"),
    ]

+    # Add user-defined custom providers from config.yaml
+    custom_providers_cfg = config.get("custom_providers") or []
+    _custom_provider_map = {}  # key → {name, base_url, api_key}
+    if isinstance(custom_providers_cfg, list):
+        for entry in custom_providers_cfg:
+            if not isinstance(entry, dict):
+                continue
+            name = entry.get("name", "").strip()
+            base_url = entry.get("base_url", "").strip()
+            if not name or not base_url:
+                continue
+            # Generate a stable key from the name
+            key = "custom:" + name.lower().replace(" ", "-")
+            short_url = base_url.replace("https://", "").replace("http://", "").rstrip("/")
+            saved_model = entry.get("model", "")
+            model_hint = f" — {saved_model}" if saved_model else ""
+            providers.append((key, f"{name} ({short_url}){model_hint}"))
+            _custom_provider_map[key] = {
+                "name": name,
+                "base_url": base_url,
+                "api_key": entry.get("api_key", ""),
+                "model": saved_model,
+            }
+
+    # Always add the manual custom endpoint option last
+    providers.append(("custom", "Custom endpoint (enter URL manually)"))
+
+    # Add removal option if there are saved custom providers
+    if _custom_provider_map:
+        providers.append(("remove-custom", "Remove a saved custom provider"))
+
    # Reorder so the active provider is at the top
    known_keys = {k for k, _ in providers}
    active_key = active if active in known_keys else "custom"
@@ -791,6 +822,10 @@ def cmd_model(args):
        _model_flow_openai_codex(config, current_model)
    elif selected_provider == "custom":
        _model_flow_custom(config)
+    elif selected_provider.startswith("custom:") and selected_provider in _custom_provider_map:
+        _model_flow_named_custom(config, _custom_provider_map[selected_provider])
+    elif selected_provider == "remove-custom":
+        _remove_custom_provider(config)
    elif selected_provider in ("zai", "kimi-coding", "minimax", "minimax-cn"):
        _model_flow_api_key_provider(config, selected_provider, current_model)

@@ -871,9 +906,11 @@ def _model_flow_openrouter(config, current_model=""):
        from hermes_cli.config import load_config, save_config
        cfg = load_config()
        model = cfg.get("model")
-        if isinstance(model, dict):
-            model["provider"] = "openrouter"
-            model["base_url"] = OPENROUTER_BASE_URL
+        if not isinstance(model, dict):
+            model = {"default": model} if model else {}
+            cfg["model"] = model
+        model["provider"] = "openrouter"
+        model["base_url"] = OPENROUTER_BASE_URL
        save_config(cfg)
        deactivate_provider()
        print(f"Default model set to: {selected} (via OpenRouter)")
@@ -1006,7 +1043,11 @@ def _model_flow_openai_codex(config, current_model=""):


 def _model_flow_custom(config):
-    """Custom endpoint: collect URL, API key, and model name."""
+    """Custom endpoint: collect URL, API key, and model name.
+
+    Automatically saves the endpoint to ``custom_providers`` in config.yaml
+    so it appears in the provider menu on subsequent runs.
+    """
    from hermes_cli.auth import _save_model_choice, deactivate_provider
    from hermes_cli.config import get_env_value, save_env_value, load_config, save_config

@@ -1038,6 +1079,8 @@ def _model_flow_custom(config):
        print(f"Invalid URL: {effective_url} (must start with http:// or https://)")
        return

+    effective_key = api_key or current_key
+
    if base_url:
        save_env_value("OPENAI_BASE_URL", base_url)
    if api_key:
@@ -1049,9 +1092,11 @@ def _model_flow_custom(config):
        # Update config and deactivate any OAuth provider
        cfg = load_config()
        model = cfg.get("model")
-        if isinstance(model, dict):
-            model["provider"] = "auto"
-            model["base_url"] = effective_url
+        if not isinstance(model, dict):
+            model = {"default": model} if model else {}
+            cfg["model"] = model
+        model["provider"] = "custom"
+        model["base_url"] = effective_url
        save_config(cfg)
        deactivate_provider()

@@ -1061,6 +1106,227 @@ def _model_flow_custom(config):
            deactivate_provider()
        print("Endpoint saved. Use `/model` in chat or `hermes model` to set a model.")

+    # Auto-save to custom_providers so it appears in the menu next time
+    _save_custom_provider(effective_url, effective_key, model_name or "")
+
+
+def _save_custom_provider(base_url, api_key="", model=""):
+    """Save a custom endpoint to custom_providers in config.yaml.
+
+    Deduplicates by base_url — if the URL already exists, updates the
+    model name but doesn't add a duplicate entry.
+    Auto-generates a display name from the URL hostname.
+    """
+    from hermes_cli.config import load_config, save_config
+
+    cfg = load_config()
+    providers = cfg.get("custom_providers") or []
+    if not isinstance(providers, list):
+        providers = []
+
+    # Check if this URL is already saved — update model if so
+    for entry in providers:
+        if isinstance(entry, dict) and entry.get("base_url", "").rstrip("/") == base_url.rstrip("/"):
+            if model and entry.get("model") != model:
+                entry["model"] = model
+                cfg["custom_providers"] = providers
+                save_config(cfg)
+            return  # already saved, updated model if needed
+
+    # Auto-generate a name from the URL
+    import re
+    clean = base_url.replace("https://", "").replace("http://", "").rstrip("/")
+    # Remove /v1 suffix for cleaner names
+    clean = re.sub(r"/v1/?$", "", clean)
+    # Use hostname:port as the name
+    name = clean.split("/")[0]
+    # Capitalize for readability
+    if "localhost" in name or "127.0.0.1" in name:
+        name = f"Local ({name})"
+    elif "runpod" in name.lower():
+        name = f"RunPod ({name})"
+    else:
+        name = name.capitalize()
+
+    entry = {"name": name, "base_url": base_url}
+    if api_key:
+        entry["api_key"] = api_key
+    if model:
+        entry["model"] = model
+
+    providers.append(entry)
+    cfg["custom_providers"] = providers
+    save_config(cfg)
+    print(f"  💾 Saved to custom providers as \"{name}\" (edit in config.yaml)")
+
+
+def _remove_custom_provider(config):
+    """Let the user remove a saved custom provider from config.yaml."""
+    from hermes_cli.config import load_config, save_config
+
+    cfg = load_config()
+    providers = cfg.get("custom_providers") or []
+    if not isinstance(providers, list) or not providers:
+        print("No custom providers configured.")
+        return
+
+    print("Remove a custom provider:\n")
+
+    choices = []
+    for entry in providers:
+        if isinstance(entry, dict):
+            name = entry.get("name", "unnamed")
+            url = entry.get("base_url", "")
+            short_url = url.replace("https://", "").replace("http://", "").rstrip("/")
+            choices.append(f"{name} ({short_url})")
+        else:
+            choices.append(str(entry))
+    choices.append("Cancel")
+
+    try:
+        from simple_term_menu import TerminalMenu
+        menu = TerminalMenu(
+            [f"  {c}" for c in choices], cursor_index=0,
+            menu_cursor="-> ", menu_cursor_style=("fg_red", "bold"),
+            menu_highlight_style=("fg_red",),
+            cycle_cursor=True, clear_screen=False,
+            title="Select provider to remove:",
+        )
+        idx = menu.show()
+        print()
+    except (ImportError, NotImplementedError):
+        for i, c in enumerate(choices, 1):
+            print(f"  {i}. {c}")
+        print()
+        try:
+            val = input(f"Choice [1-{len(choices)}]: ").strip()
+            idx = int(val) - 1 if val else None
+        except (ValueError, KeyboardInterrupt, EOFError):
+            idx = None
+
+    if idx is None or idx >= len(providers):
+        print("No change.")
+        return
+
+    removed = providers.pop(idx)
+    cfg["custom_providers"] = providers
+    save_config(cfg)
+    removed_name = removed.get("name", "unnamed") if isinstance(removed, dict) else str(removed)
+    print(f"✅ Removed \"{removed_name}\" from custom providers.")
+
+
+def _model_flow_named_custom(config, provider_info):
+    """Handle a named custom provider from config.yaml custom_providers list.
+
+    If the entry has a saved model name, activates it immediately.
+    Otherwise probes the endpoint's /models API to let the user pick one.
+    """
+    from hermes_cli.auth import _save_model_choice, deactivate_provider
+    from hermes_cli.config import save_env_value, load_config, save_config
+    from hermes_cli.models import fetch_api_models
+
+    name = provider_info["name"]
+    base_url = provider_info["base_url"]
+    api_key = provider_info.get("api_key", "")
+    saved_model = provider_info.get("model", "")
+
+    # If a model is saved, just activate immediately — no probing needed
+    if saved_model:
+        save_env_value("OPENAI_BASE_URL", base_url)
+        if api_key:
+            save_env_value("OPENAI_API_KEY", api_key)
+        _save_model_choice(saved_model)
+
+        cfg = load_config()
+        model = cfg.get("model")
+        if not isinstance(model, dict):
+            model = {"default": model} if model else {}
+            cfg["model"] = model
+        model["provider"] = "custom"
+        model["base_url"] = base_url
+        save_config(cfg)
+        deactivate_provider()
+
+        print(f"✅ Switched to: {saved_model}")
+        print(f"   Provider: {name} ({base_url})")
+        return
+
+    # No saved model — probe endpoint and let user pick
+    print(f"  Provider: {name}")
+    print(f"  URL:      {base_url}")
+    print()
+    print("No model saved for this provider. Fetching available models...")
+    models = fetch_api_models(api_key, base_url, timeout=8.0)
+
+    if models:
+        print(f"Found {len(models)} model(s):\n")
+        try:
+            from simple_term_menu import TerminalMenu
+            menu_items = [f"  {m}" for m in models] + ["  Cancel"]
+            menu = TerminalMenu(
+                menu_items, cursor_index=0,
+                menu_cursor="-> ", menu_cursor_style=("fg_green", "bold"),
+                menu_highlight_style=("fg_green",),
+                cycle_cursor=True, clear_screen=False,
+                title=f"Select model from {name}:",
+            )
+            idx = menu.show()
+            print()
+            if idx is None or idx >= len(models):
+                print("Cancelled.")
+                return
+            model_name = models[idx]
+        except (ImportError, NotImplementedError):
+            for i, m in enumerate(models, 1):
+                print(f"  {i}. {m}")
+            print(f"  {len(models) + 1}. Cancel")
+            print()
+            try:
+                val = input(f"Choice [1-{len(models) + 1}]: ").strip()
+                if not val:
+                    print("Cancelled.")
+                    return
+                idx = int(val) - 1
+                if idx < 0 or idx >= len(models):
+                    print("Cancelled.")
+                    return
+                model_name = models[idx]
+            except (ValueError, KeyboardInterrupt, EOFError):
+                print("\nCancelled.")
+                return
+    else:
+        print("Could not fetch models from endpoint. Enter model name manually.")
+        try:
+            model_name = input("Model name: ").strip()
+        except (KeyboardInterrupt, EOFError):
+            print("\nCancelled.")
+            return
+        if not model_name:
+            print("No model specified. Cancelled.")
+            return
+
+    # Activate and save the model to the custom_providers entry
+    save_env_value("OPENAI_BASE_URL", base_url)
+    if api_key:
+        save_env_value("OPENAI_API_KEY", api_key)
+    _save_model_choice(model_name)
+
+    cfg = load_config()
+    model = cfg.get("model")
+    if not isinstance(model, dict):
+        model = {"default": model} if model else {}
+        cfg["model"] = model
+    model["provider"] = "custom"
+    model["base_url"] = base_url
+    save_config(cfg)
+    deactivate_provider()
+
+    # Save model name to the custom_providers entry for next time
+    _save_custom_provider(base_url, api_key, model_name)
+
+    print(f"\n✅ Model set to: {model_name}")
+    print(f"   Provider: {name} ({base_url})")
+

 # Curated model lists for direct API-key providers
 _PROVIDER_MODELS = {
@@ -1162,9 +1428,11 @@ def _model_flow_api_key_provider(config, provider_id, current_model=""):
        # Update config with provider and base URL
        cfg = load_config()
        model = cfg.get("model")
-        if isinstance(model, dict):
-            model["provider"] = provider_id
-            model["base_url"] = effective_base
+        if not isinstance(model, dict):
+            model = {"default": model} if model else {}
+            cfg["model"] = model
+        model["provider"] = provider_id
+        model["base_url"] = effective_base
        save_config(cfg)
        deactivate_provider()

@@ -1520,6 +1788,44 @@ def cmd_update(args):
            sys.exit(1)


+def _coalesce_session_name_args(argv: list) -> list:
+    """Join unquoted multi-word session names after -c/--continue and -r/--resume.
+
+    When a user types ``hermes -c Pokemon Agent Dev`` without quoting the
+    session name, argparse sees three separate tokens.  This function merges
+    them into a single argument so argparse receives
+    ``['-c', 'Pokemon Agent Dev']`` instead.
+
+    Tokens are collected after the flag until we hit another flag (``-*``)
+    or a known top-level subcommand.
+    """
+    _SUBCOMMANDS = {
+        "chat", "model", "gateway", "setup", "whatsapp", "login", "logout",
+        "status", "cron", "doctor", "config", "pairing", "skills", "tools",
+        "sessions", "insights", "version", "update", "uninstall",
+    }
+    _SESSION_FLAGS = {"-c", "--continue", "-r", "--resume"}
+
+    result = []
+    i = 0
+    while i < len(argv):
+        token = argv[i]
+        if token in _SESSION_FLAGS:
+            result.append(token)
+            i += 1
+            # Collect subsequent non-flag, non-subcommand tokens as one name
+            parts: list = []
+            while i < len(argv) and not argv[i].startswith("-") and argv[i] not in _SUBCOMMANDS:
+                parts.append(argv[i])
+                i += 1
+            if parts:
+                result.append(" ".join(parts))
+        else:
+            result.append(token)
+            i += 1
+    return result
+
+
 def main():
    """Main entry point for hermes CLI."""
    parser = argparse.ArgumentParser(
@@ -1632,6 +1938,12 @@ For more help on a command:
        default=False,
        help="Run in an isolated git worktree (for parallel agents on the same repo)"
    )
+    chat_parser.add_argument(
+        "--checkpoints",
+        action="store_true",
+        default=False,
+        help="Enable filesystem checkpoints before destructive file operations (use /rollback to restore)"
+    )
    chat_parser.set_defaults(func=cmd_chat)

    # =========================================================================
@@ -2099,12 +2411,12 @@ For more help on a command:
                if not data:
                    print(f"Session '{args.session_id}' not found.")
                    return
-                with open(args.output, "w") as f:
+                with open(args.output, "w", encoding="utf-8") as f:
                    f.write(_json.dumps(data, ensure_ascii=False) + "\n")
                print(f"Exported 1 session to {args.output}")
            else:
                sessions = db.export_all(source=args.source)
-                with open(args.output, "w") as f:
+                with open(args.output, "w", encoding="utf-8") as f:
                    for s in sessions:
                        f.write(_json.dumps(s, ensure_ascii=False) + "\n")
                print(f"Exported {len(sessions)} sessions to {args.output}")
@@ -2258,7 +2570,11 @@ For more help on a command:
    # =========================================================================
    # Parse and execute
    # =========================================================================
-    args = parser.parse_args()
+    # Pre-process argv so unquoted multi-word session names after -c / -r
+    # are merged into a single token before argparse sees them.
+    # e.g. ``hermes -c Pokemon Agent Dev`` → ``hermes -c 'Pokemon Agent Dev'``
+    _processed_argv = _coalesce_session_name_args(sys.argv[1:])
+    args = parser.parse_args(_processed_argv)
    
    # Handle --version flag
    if args.version:
--- a/hermes_cli/models.py
+++ b/hermes_cli/models.py
@@ -63,7 +63,7 @@ _PROVIDER_LABELS = {
    "kimi-coding": "Kimi / Moonshot",
    "minimax": "MiniMax",
    "minimax-cn": "MiniMax (China)",
-    "custom": "custom endpoint",
+    "custom": "Custom endpoint",
 }

 _PROVIDER_ALIASES = {
--- a/hermes_cli/runtime_provider.py
+++ b/hermes_cli/runtime_provider.py
@@ -66,9 +66,14 @@ def _resolve_openrouter_runtime(
            if not cfg_provider or cfg_provider == "auto":
                use_config_base_url = True

+    # When the user explicitly requested the openrouter provider, skip
+    # OPENAI_BASE_URL — it typically points to a custom / non-OpenRouter
+    # endpoint and would prevent switching back to OpenRouter (#874).
+    skip_openai_base = requested_norm == "openrouter"
+
    base_url = (
        (explicit_base_url or "").strip()
-        or env_openai_base_url
+        or ("" if skip_openai_base else env_openai_base_url)
        or (cfg_base_url.strip() if use_config_base_url else "")
        or env_openrouter_base_url
        or OPENROUTER_BASE_URL
--- a/hermes_cli/setup.py
+++ b/hermes_cli/setup.py
@@ -516,7 +516,8 @@ def setup_model_provider(config: dict):
        keep_label = None  # No provider configured — don't show "Keep current"

    provider_choices = [
-        "Login with Nous Portal (Nous Research subscription)",
+        "Nous Portal API key (direct API key access)",
+        "Login with Nous Portal (Nous Research subscription — OAuth)",
        "Login with OpenAI Codex",
        "OpenRouter API key (100+ models, pay-per-use)",
        "Custom OpenAI-compatible endpoint (self-hosted / VLLM / etc.)",
@@ -529,7 +530,7 @@ def setup_model_provider(config: dict):
        provider_choices.append(keep_label)
    
    # Default to "Keep current" if a provider exists, otherwise OpenRouter (most common)
-    default_provider = len(provider_choices) - 1 if has_any_provider else 2
+    default_provider = len(provider_choices) - 1 if has_any_provider else 3
    
    if not has_any_provider:
        print_warning("An inference provider is required for Hermes to work.")
@@ -541,7 +542,37 @@ def setup_model_provider(config: dict):
    selected_provider = None  # "nous", "openai-codex", "openrouter", "custom", or None (keep)
    nous_models = []  # populated if Nous login succeeds

-    if provider_idx == 0:  # Nous Portal
+    if provider_idx == 0:  # Nous Portal API Key (direct)
+        selected_provider = "nous-api"
+        print()
+        print_header("Nous Portal API Key")
+        print_info("Use a Nous Portal API key for direct access to Nous inference.")
+        print_info("Get your API key at: https://portal.nousresearch.com")
+        print()
+
+        existing_key = get_env_value("NOUS_API_KEY")
+        if existing_key:
+            print_info(f"Current: {existing_key[:8]}... (configured)")
+            if prompt_yes_no("Update Nous API key?", False):
+                api_key = prompt("  Nous API key", password=True)
+                if api_key:
+                    save_env_value("NOUS_API_KEY", api_key)
+                    print_success("Nous API key updated")
+        else:
+            api_key = prompt("  Nous API key", password=True)
+            if api_key:
+                save_env_value("NOUS_API_KEY", api_key)
+                print_success("Nous API key saved")
+            else:
+                print_warning("Skipped - agent won't work without an API key")
+
+        # Clear custom endpoint vars if switching
+        if existing_custom:
+            save_env_value("OPENAI_BASE_URL", "")
+            save_env_value("OPENAI_API_KEY", "")
+        _update_config_for_provider("nous-api", "https://inference-api.nousresearch.com/v1")
+
+    elif provider_idx == 1:  # Nous Portal
        selected_provider = "nous"
        print()
        print_header("Nous Portal Login")
@@ -581,7 +612,7 @@ def setup_model_provider(config: dict):
            print_info("You can try again later with: hermes model")
            selected_provider = None

-    elif provider_idx == 1:  # OpenAI Codex
+    elif provider_idx == 2:  # OpenAI Codex
        selected_provider = "openai-codex"
        print()
        print_header("OpenAI Codex Login")
@@ -605,7 +636,7 @@ def setup_model_provider(config: dict):
            print_info("You can try again later with: hermes model")
            selected_provider = None

-    elif provider_idx == 2:  # OpenRouter
+    elif provider_idx == 3:  # OpenRouter
        selected_provider = "openrouter"
        print()
        print_header("OpenRouter API Key")
@@ -632,7 +663,30 @@ def setup_model_provider(config: dict):
            save_env_value("OPENAI_BASE_URL", "")
            save_env_value("OPENAI_API_KEY", "")

-    elif provider_idx == 3:  # Custom endpoint
+        # Update config.yaml and deactivate any OAuth provider so the
+        # resolver doesn't keep returning the old provider (e.g. Codex).
+        try:
+            from hermes_cli.auth import deactivate_provider
+            deactivate_provider()
+        except Exception:
+            pass
+        import yaml
+        config_path = Path(os.environ.get("HERMES_HOME", Path.home() / ".hermes")) / "config.yaml"
+        try:
+            disk_cfg = {}
+            if config_path.exists():
+                disk_cfg = yaml.safe_load(config_path.read_text()) or {}
+            model_section = disk_cfg.get("model", {})
+            if isinstance(model_section, str):
+                model_section = {"default": model_section}
+            model_section["provider"] = "openrouter"
+            model_section.pop("base_url", None)  # OpenRouter uses default URL
+            disk_cfg["model"] = model_section
+            config_path.write_text(yaml.safe_dump(disk_cfg, sort_keys=False))
+        except Exception as e:
+            logger.debug("Could not save provider to config.yaml: %s", e)
+
+    elif provider_idx == 4:  # Custom endpoint
        selected_provider = "custom"
        print()
        print_header("Custom OpenAI-Compatible Endpoint")
@@ -659,9 +713,31 @@ def setup_model_provider(config: dict):
        if model_name:
            config['model'] = model_name
            save_env_value("LLM_MODEL", model_name)
+
+        # Save provider and base_url to config.yaml so the gateway and CLI
+        # both resolve the correct provider without relying on env-var heuristics.
+        if base_url:
+            import yaml
+            config_path = Path(os.environ.get("HERMES_HOME", Path.home() / ".hermes")) / "config.yaml"
+            try:
+                disk_cfg = {}
+                if config_path.exists():
+                    disk_cfg = yaml.safe_load(config_path.read_text()) or {}
+                model_section = disk_cfg.get("model", {})
+                if isinstance(model_section, str):
+                    model_section = {"default": model_section}
+                model_section["provider"] = "custom"
+                model_section["base_url"] = base_url.rstrip("/")
+                if model_name:
+                    model_section["default"] = model_name
+                disk_cfg["model"] = model_section
+                config_path.write_text(yaml.safe_dump(disk_cfg, sort_keys=False))
+            except Exception as e:
+                logger.debug("Could not save provider to config.yaml: %s", e)
+
        print_success("Custom endpoint configured")

-    elif provider_idx == 4:  # Z.AI / GLM
+    elif provider_idx == 5:  # Z.AI / GLM
        selected_provider = "zai"
        print()
        print_header("Z.AI / GLM API Key")
@@ -715,7 +791,7 @@ def setup_model_provider(config: dict):
            save_env_value("OPENAI_API_KEY", "")
        _update_config_for_provider("zai", zai_base_url)

-    elif provider_idx == 5:  # Kimi / Moonshot
+    elif provider_idx == 6:  # Kimi / Moonshot
        selected_provider = "kimi-coding"
        print()
        print_header("Kimi / Moonshot API Key")
@@ -747,7 +823,7 @@ def setup_model_provider(config: dict):
            save_env_value("OPENAI_API_KEY", "")
        _update_config_for_provider("kimi-coding", pconfig.inference_base_url)

-    elif provider_idx == 6:  # MiniMax
+    elif provider_idx == 7:  # MiniMax
        selected_provider = "minimax"
        print()
        print_header("MiniMax API Key")
@@ -779,7 +855,7 @@ def setup_model_provider(config: dict):
            save_env_value("OPENAI_API_KEY", "")
        _update_config_for_provider("minimax", pconfig.inference_base_url)

-    elif provider_idx == 7:  # MiniMax China
+    elif provider_idx == 8:  # MiniMax China
        selected_provider = "minimax-cn"
        print()
        print_header("MiniMax China API Key")
@@ -811,12 +887,12 @@ def setup_model_provider(config: dict):
            save_env_value("OPENAI_API_KEY", "")
        _update_config_for_provider("minimax-cn", pconfig.inference_base_url)

-    # else: provider_idx == 8 (Keep current) — only shown when a provider already exists
+    # else: provider_idx == 9 (Keep current) — only shown when a provider already exists

    # ── OpenRouter API Key for tools (if not already set) ──
    # Tools (vision, web, MoA) use OpenRouter independently of the main provider.
    # Prompt for OpenRouter key if not set and a non-OpenRouter provider was chosen.
-    if selected_provider in ("nous", "openai-codex", "custom", "zai", "kimi-coding", "minimax", "minimax-cn") and not get_env_value("OPENROUTER_API_KEY"):
+    if selected_provider in ("nous", "nous-api", "openai-codex", "custom", "zai", "kimi-coding", "minimax", "minimax-cn") and not get_env_value("OPENROUTER_API_KEY"):
        print()
        print_header("OpenRouter API Key (for tools)")
        print_info("Tools like vision analysis, web search, and MoA use OpenRouter")
@@ -869,6 +945,14 @@ def setup_model_provider(config: dict):
            if custom:
                config['model'] = custom
                save_env_value("LLM_MODEL", custom)
+        elif selected_provider == "nous-api":
+            # Nous API key provider — prompt for model manually
+            print_info("Enter a model name available on Nous inference API.")
+            print_info("Examples: anthropic/claude-opus-4.6, deepseek/deepseek-r1")
+            custom = prompt(f"  Model name (Enter to keep '{current_model}')")
+            if custom:
+                config['model'] = custom
+                save_env_value("LLM_MODEL", custom)
        elif selected_provider == "openai-codex":
            from hermes_cli.codex_models import get_codex_model_ids
            codex_models = get_codex_model_ids()
@@ -1264,7 +1348,7 @@ def setup_agent_settings(config: dict):
    # ── Max Iterations ──
    print_header("Agent Settings")

-    current_max = get_env_value('HERMES_MAX_ITERATIONS') or '60'
+    current_max = get_env_value('HERMES_MAX_ITERATIONS') or str(config.get('agent', {}).get('max_turns', 90))
    print_info("Maximum tool-calling iterations per conversation.")
    print_info("Higher = more complex tasks, but costs more tokens.")
    print_info("Recommended: 30-60 for most tasks, 100+ for open exploration.")
@@ -1274,7 +1358,8 @@ def setup_agent_settings(config: dict):
        max_iter = int(max_iter_str)
        if max_iter > 0:
            save_env_value("HERMES_MAX_ITERATIONS", str(max_iter))
-            config['max_turns'] = max_iter
+            config.setdefault('agent', {})['max_turns'] = max_iter
+            config.pop('max_turns', None)
            print_success(f"Max iterations set to {max_iter}")
    except ValueError:
        print_warning("Invalid number, keeping current value")
@@ -1527,10 +1612,22 @@ def setup_gateway(config: dict):
    
    if not existing_slack and prompt_yes_no("Set up Slack bot?", False):
        print_info("Steps to create a Slack app:")
-        print_info("   1. Go to https://api.slack.com/apps → Create New App")
-        print_info("   2. Enable Socket Mode: App Settings → Socket Mode → Enable")
-        print_info("   3. Bot Token: OAuth & Permissions → Install to Workspace")
-        print_info("   4. App Token: Basic Information → App-Level Tokens → Generate")
+        print_info("   1. Go to https://api.slack.com/apps → Create New App (from scratch)")
+        print_info("   2. Enable Socket Mode: Settings → Socket Mode → Enable")
+        print_info("      • Create an App-Level Token with 'connections:write' scope")
+        print_info("   3. Add Bot Token Scopes: Features → OAuth & Permissions")
+        print_info("      Required scopes: chat:write, app_mentions:read,")
+        print_info("      channels:history, channels:read, groups:history,")
+        print_info("      im:history, im:read, im:write, users:read, files:write")
+        print_info("   4. Subscribe to Events: Features → Event Subscriptions → Enable")
+        print_info("      Required events: message.im, message.channels,")
+        print_info("      message.groups, app_mention")
+        print_warning("   ⚠ Without message.channels/message.groups events,")
+        print_warning("     the bot will ONLY work in DMs, not channels!")
+        print_info("   5. Install to Workspace: Settings → Install App")
+        print_info("   6. After installing, invite the bot to channels: /invite @YourBot")
+        print()
+        print_info("   Full guide: https://hermes-agent.ai/docs/user-guide/messaging/slack")
        print()
        bot_token = prompt("Slack Bot Token (xoxb-...)", password=True)
        if bot_token:
@@ -1542,7 +1639,7 @@ def setup_gateway(config: dict):
            
            print()
            print_info("🔒 Security: Restrict who can use your bot")
-            print_info("   Find Slack user IDs in your profile or via the Slack API")
+            print_info("   To find a Member ID: click a user's name → View full profile → ⋮ → Copy member ID")
            print()
            allowed_users = prompt("Allowed user IDs (comma-separated, leave empty for open access)")
            if allowed_users:
@@ -1660,14 +1757,18 @@ def setup_gateway(config: dict):
 # Section 5: Tool Configuration (delegates to unified tools_config.py)
 # =============================================================================

-def setup_tools(config: dict):
+def setup_tools(config: dict, first_install: bool = False):
    """Configure tools — delegates to the unified tools_command() in tools_config.py.
    
    Both `hermes setup tools` and `hermes tools` use the same flow:
    platform selection → toolset toggles → provider/API key configuration.
+    
+    Args:
+        first_install: When True, uses the simplified first-install flow
+            (no platform menu, prompts for all unconfigured API keys).
    """
    from hermes_cli.tools_config import tools_command
-    tools_command()
+    tools_command(first_install=first_install, config=config)


 # =============================================================================
@@ -1820,7 +1921,7 @@ def run_setup_wizard(args):
    setup_gateway(config)

    # Section 5: Tools
-    setup_tools(config)
+    setup_tools(config, first_install=not is_existing)

    # Save and show summary
    save_config(config)
--- a/hermes_cli/skin_engine.py
+++ b/hermes_cli/skin_engine.py
@@ -0,0 +1,630 @@
+"""Hermes CLI skin/theme engine.
+
+A data-driven skin system that lets users customize the CLI's visual appearance.
+Skins are defined as YAML files in ~/.hermes/skins/ or as built-in presets.
+No code changes are needed to add a new skin.
+
+SKIN YAML SCHEMA
+================
+
+All fields are optional. Missing values inherit from the ``default`` skin.
+
+.. code-block:: yaml
+
+    # Required: skin identity
+    name: mytheme                         # Unique skin name (lowercase, hyphens ok)
+    description: Short description        # Shown in /skin listing
+
+    # Colors: hex values for Rich markup (banner, UI, response box)
+    colors:
+      banner_border: "#CD7F32"            # Panel border color
+      banner_title: "#FFD700"             # Panel title text color
+      banner_accent: "#FFBF00"            # Section headers (Available Tools, etc.)
+      banner_dim: "#B8860B"               # Dim/muted text (separators, labels)
+      banner_text: "#FFF8DC"              # Body text (tool names, skill names)
+      ui_accent: "#FFBF00"               # General UI accent
+      ui_label: "#4dd0e1"                # UI labels
+      ui_ok: "#4caf50"                   # Success indicators
+      ui_error: "#ef5350"                # Error indicators
+      ui_warn: "#ffa726"                 # Warning indicators
+      prompt: "#FFF8DC"                  # Prompt text color
+      input_rule: "#CD7F32"              # Input area horizontal rule
+      response_border: "#FFD700"         # Response box border (ANSI)
+      session_label: "#DAA520"           # Session label color
+      session_border: "#8B8682"          # Session ID dim color
+
+    # Spinner: customize the animated spinner during API calls
+    spinner:
+      waiting_faces:                      # Faces shown while waiting for API
+        - "(⚔)"
+        - "(⛨)"
+      thinking_faces:                     # Faces shown during reasoning
+        - "(⌁)"
+        - "(<>)"
+      thinking_verbs:                     # Verbs for spinner messages
+        - "forging"
+        - "plotting"
+      wings:                              # Optional left/right spinner decorations
+        - ["⟪⚔", "⚔⟫"]                  # Each entry is [left, right] pair
+        - ["⟪▲", "▲⟫"]
+
+    # Branding: text strings used throughout the CLI
+    branding:
+      agent_name: "Hermes Agent"          # Banner title, status display
+      welcome: "Welcome message"          # Shown at CLI startup
+      goodbye: "Goodbye! ⚕"              # Shown on exit
+      response_label: " ⚕ Hermes "       # Response box header label
+      prompt_symbol: "❯ "                # Input prompt symbol
+      help_header: "(^_^)? Commands"      # /help header text
+
+    # Tool prefix: character for tool output lines (default: ┊)
+    tool_prefix: "┊"
+
+USAGE
+=====
+
+.. code-block:: python
+
+    from hermes_cli.skin_engine import get_active_skin, list_skins, set_active_skin
+
+    skin = get_active_skin()
+    print(skin.colors["banner_title"])    # "#FFD700"
+    print(skin.get_branding("agent_name"))  # "Hermes Agent"
+
+    set_active_skin("ares")               # Switch to built-in ares skin
+    set_active_skin("mytheme")            # Switch to user skin from ~/.hermes/skins/
+
+BUILT-IN SKINS
+==============
+
+- ``default`` — Classic Hermes gold/kawaii (the current look)
+- ``ares``    — Crimson/bronze war-god theme with custom spinner wings
+- ``mono``    — Clean grayscale monochrome
+- ``slate``   — Cool blue developer-focused theme
+
+USER SKINS
+==========
+
+Drop a YAML file in ``~/.hermes/skins/<name>.yaml`` following the schema above.
+Activate with ``/skin <name>`` in the CLI or ``display.skin: <name>`` in config.yaml.
+"""
+
+import logging
+import os
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Tuple
+
+logger = logging.getLogger(__name__)
+
+
+# =============================================================================
+# Skin data structure
+# =============================================================================
+
+@dataclass
+class SkinConfig:
+    """Complete skin configuration."""
+    name: str
+    description: str = ""
+    colors: Dict[str, str] = field(default_factory=dict)
+    spinner: Dict[str, Any] = field(default_factory=dict)
+    branding: Dict[str, str] = field(default_factory=dict)
+    tool_prefix: str = "┊"
+    banner_logo: str = ""    # Rich-markup ASCII art logo (replaces HERMES_AGENT_LOGO)
+    banner_hero: str = ""    # Rich-markup hero art (replaces HERMES_CADUCEUS)
+
+    def get_color(self, key: str, fallback: str = "") -> str:
+        """Get a color value with fallback."""
+        return self.colors.get(key, fallback)
+
+    def get_spinner_list(self, key: str) -> List[str]:
+        """Get a spinner list (faces, verbs, etc.)."""
+        return self.spinner.get(key, [])
+
+    def get_spinner_wings(self) -> List[Tuple[str, str]]:
+        """Get spinner wing pairs, or empty list if none."""
+        raw = self.spinner.get("wings", [])
+        result = []
+        for pair in raw:
+            if isinstance(pair, (list, tuple)) and len(pair) == 2:
+                result.append((str(pair[0]), str(pair[1])))
+        return result
+
+    def get_branding(self, key: str, fallback: str = "") -> str:
+        """Get a branding value with fallback."""
+        return self.branding.get(key, fallback)
+
+
+# =============================================================================
+# Built-in skin definitions
+# =============================================================================
+
+_BUILTIN_SKINS: Dict[str, Dict[str, Any]] = {
+    "default": {
+        "name": "default",
+        "description": "Classic Hermes — gold and kawaii",
+        "colors": {
+            "banner_border": "#CD7F32",
+            "banner_title": "#FFD700",
+            "banner_accent": "#FFBF00",
+            "banner_dim": "#B8860B",
+            "banner_text": "#FFF8DC",
+            "ui_accent": "#FFBF00",
+            "ui_label": "#4dd0e1",
+            "ui_ok": "#4caf50",
+            "ui_error": "#ef5350",
+            "ui_warn": "#ffa726",
+            "prompt": "#FFF8DC",
+            "input_rule": "#CD7F32",
+            "response_border": "#FFD700",
+            "session_label": "#DAA520",
+            "session_border": "#8B8682",
+        },
+        "spinner": {
+            # Empty = use hardcoded defaults in display.py
+        },
+        "branding": {
+            "agent_name": "Hermes Agent",
+            "welcome": "Welcome to Hermes Agent! Type your message or /help for commands.",
+            "goodbye": "Goodbye! ⚕",
+            "response_label": " ⚕ Hermes ",
+            "prompt_symbol": "❯ ",
+            "help_header": "(^_^)? Available Commands",
+        },
+        "tool_prefix": "┊",
+    },
+    "ares": {
+        "name": "ares",
+        "description": "War-god theme — crimson and bronze",
+        "colors": {
+            "banner_border": "#9F1C1C",
+            "banner_title": "#C7A96B",
+            "banner_accent": "#DD4A3A",
+            "banner_dim": "#6B1717",
+            "banner_text": "#F1E6CF",
+            "ui_accent": "#DD4A3A",
+            "ui_label": "#C7A96B",
+            "ui_ok": "#4caf50",
+            "ui_error": "#ef5350",
+            "ui_warn": "#ffa726",
+            "prompt": "#F1E6CF",
+            "input_rule": "#9F1C1C",
+            "response_border": "#C7A96B",
+            "session_label": "#C7A96B",
+            "session_border": "#6E584B",
+        },
+        "spinner": {
+            "waiting_faces": ["(⚔)", "(⛨)", "(▲)", "(<>)", "(/)"],
+            "thinking_faces": ["(⚔)", "(⛨)", "(▲)", "(⌁)", "(<>)"],
+            "thinking_verbs": [
+                "forging", "marching", "sizing the field", "holding the line",
+                "hammering plans", "tempering steel", "plotting impact", "raising the shield",
+            ],
+            "wings": [
+                ["⟪⚔", "⚔⟫"],
+                ["⟪▲", "▲⟫"],
+                ["⟪╸", "╺⟫"],
+                ["⟪⛨", "⛨⟫"],
+            ],
+        },
+        "branding": {
+            "agent_name": "Ares Agent",
+            "welcome": "Welcome to Ares Agent! Type your message or /help for commands.",
+            "goodbye": "Farewell, warrior! ⚔",
+            "response_label": " ⚔ Ares ",
+            "prompt_symbol": "⚔ ❯ ",
+            "help_header": "(⚔) Available Commands",
+        },
+        "tool_prefix": "╎",
+        "banner_logo": """[bold #A3261F] █████╗ ██████╗ ███████╗███████╗       █████╗  ██████╗ ███████╗███╗   ██╗████████╗[/]
+[bold #B73122]██╔══██╗██╔══██╗██╔════╝██╔════╝      ██╔══██╗██╔════╝ ██╔════╝████╗  ██║╚══██╔══╝[/]
+[#C93C24]███████║██████╔╝█████╗  ███████╗█████╗███████║██║  ███╗█████╗  ██╔██╗ ██║   ██║[/]
+[#D84A28]██╔══██║██╔══██╗██╔══╝  ╚════██║╚════╝██╔══██║██║   ██║██╔══╝  ██║╚██╗██║   ██║[/]
+[#E15A2D]██║  ██║██║  ██║███████╗███████║      ██║  ██║╚██████╔╝███████╗██║ ╚████║   ██║[/]
+[#EB6C32]╚═╝  ╚═╝╚═╝  ╚═╝╚══════╝╚══════╝      ╚═╝  ╚═╝ ╚═════╝ ╚══════╝╚═╝  ╚═══╝   ╚═╝[/]""",
+        "banner_hero": """[#9F1C1C]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣤⣤⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#9F1C1C]⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣴⣿⠟⠻⣿⣦⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#C7A96B]⠀⠀⠀⠀⠀⠀⠀⣠⣾⡿⠋⠀⠀⠀⠙⢿⣷⣄⠀⠀⠀⠀⠀⠀⠀[/]
+[#C7A96B]⠀⠀⠀⠀⠀⢀⣾⡿⠋⠀⠀⢠⡄⠀⠀⠙⢿⣷⡀⠀⠀⠀⠀⠀[/]
+[#DD4A3A]⠀⠀⠀⠀⣰⣿⠟⠀⠀⠀⣰⣿⣿⣆⠀⠀⠀⠻⣿⣆⠀⠀⠀⠀[/]
+[#DD4A3A]⠀⠀⠀⢰⣿⠏⠀⠀⢀⣾⡿⠉⢿⣷⡀⠀⠀⠹⣿⡆⠀⠀⠀[/]
+[#9F1C1C]⠀⠀⠀⣿⡟⠀⠀⣠⣿⠟⠀⠀⠀⠻⣿⣄⠀⠀⢻⣿⠀⠀⠀[/]
+[#9F1C1C]⠀⠀⠀⣿⡇⠀⠀⠙⠋⠀⠀⚔⠀⠀⠙⠋⠀⠀⢸⣿⠀⠀⠀[/]
+[#6B1717]⠀⠀⠀⢿⣧⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣼⡿⠀⠀⠀[/]
+[#6B1717]⠀⠀⠀⠘⢿⣷⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⣾⡿⠃⠀⠀⠀[/]
+[#C7A96B]⠀⠀⠀⠀⠈⠻⣿⣷⣦⣤⣀⣀⣤⣤⣶⣿⠿⠋⠀⠀⠀⠀[/]
+[#C7A96B]⠀⠀⠀⠀⠀⠀⠀⠉⠛⠿⠿⠿⠿⠛⠉⠀⠀⠀⠀⠀⠀⠀[/]
+[#DD4A3A]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⚔⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[dim #6B1717]⠀⠀⠀⠀⠀⠀⠀⠀war god online⠀⠀⠀⠀⠀⠀⠀⠀[/]""",
+    },
+    "mono": {
+        "name": "mono",
+        "description": "Monochrome — clean grayscale",
+        "colors": {
+            "banner_border": "#555555",
+            "banner_title": "#e6edf3",
+            "banner_accent": "#aaaaaa",
+            "banner_dim": "#444444",
+            "banner_text": "#c9d1d9",
+            "ui_accent": "#aaaaaa",
+            "ui_label": "#888888",
+            "ui_ok": "#888888",
+            "ui_error": "#cccccc",
+            "ui_warn": "#999999",
+            "prompt": "#c9d1d9",
+            "input_rule": "#444444",
+            "response_border": "#aaaaaa",
+            "session_label": "#888888",
+            "session_border": "#555555",
+        },
+        "spinner": {},
+        "branding": {
+            "agent_name": "Hermes Agent",
+            "welcome": "Welcome to Hermes Agent! Type your message or /help for commands.",
+            "goodbye": "Goodbye! ⚕",
+            "response_label": " ⚕ Hermes ",
+            "prompt_symbol": "❯ ",
+            "help_header": "[?] Available Commands",
+        },
+        "tool_prefix": "┊",
+    },
+    "slate": {
+        "name": "slate",
+        "description": "Cool blue — developer-focused",
+        "colors": {
+            "banner_border": "#4169e1",
+            "banner_title": "#7eb8f6",
+            "banner_accent": "#8EA8FF",
+            "banner_dim": "#4b5563",
+            "banner_text": "#c9d1d9",
+            "ui_accent": "#7eb8f6",
+            "ui_label": "#8EA8FF",
+            "ui_ok": "#63D0A6",
+            "ui_error": "#F7A072",
+            "ui_warn": "#e6a855",
+            "prompt": "#c9d1d9",
+            "input_rule": "#4169e1",
+            "response_border": "#7eb8f6",
+            "session_label": "#7eb8f6",
+            "session_border": "#4b5563",
+        },
+        "spinner": {},
+        "branding": {
+            "agent_name": "Hermes Agent",
+            "welcome": "Welcome to Hermes Agent! Type your message or /help for commands.",
+            "goodbye": "Goodbye! ⚕",
+            "response_label": " ⚕ Hermes ",
+            "prompt_symbol": "❯ ",
+            "help_header": "(^_^)? Available Commands",
+        },
+        "tool_prefix": "┊",
+    },
+    "poseidon": {
+        "name": "poseidon",
+        "description": "Ocean-god theme — deep blue and seafoam",
+        "colors": {
+            "banner_border": "#2A6FB9",
+            "banner_title": "#A9DFFF",
+            "banner_accent": "#5DB8F5",
+            "banner_dim": "#153C73",
+            "banner_text": "#EAF7FF",
+            "ui_accent": "#5DB8F5",
+            "ui_label": "#A9DFFF",
+            "ui_ok": "#4caf50",
+            "ui_error": "#ef5350",
+            "ui_warn": "#ffa726",
+            "prompt": "#EAF7FF",
+            "input_rule": "#2A6FB9",
+            "response_border": "#5DB8F5",
+            "session_label": "#A9DFFF",
+            "session_border": "#496884",
+        },
+        "spinner": {
+            "waiting_faces": ["(≈)", "(Ψ)", "(∿)", "(◌)", "(◠)"],
+            "thinking_faces": ["(Ψ)", "(∿)", "(≈)", "(⌁)", "(◌)"],
+            "thinking_verbs": [
+                "charting currents", "sounding the depth", "reading foam lines",
+                "steering the trident", "tracking undertow", "plotting sea lanes",
+                "calling the swell", "measuring pressure",
+            ],
+            "wings": [
+                ["⟪≈", "≈⟫"],
+                ["⟪Ψ", "Ψ⟫"],
+                ["⟪∿", "∿⟫"],
+                ["⟪◌", "◌⟫"],
+            ],
+        },
+        "branding": {
+            "agent_name": "Poseidon Agent",
+            "welcome": "Welcome to Poseidon Agent! Type your message or /help for commands.",
+            "goodbye": "Fair winds! Ψ",
+            "response_label": " Ψ Poseidon ",
+            "prompt_symbol": "Ψ ❯ ",
+            "help_header": "(Ψ) Available Commands",
+        },
+        "tool_prefix": "│",
+        "banner_logo": """[bold #B8E8FF]██████╗  ██████╗ ███████╗██╗██████╗ ███████╗ ██████╗ ███╗   ██╗       █████╗  ██████╗ ███████╗███╗   ██╗████████╗[/]
+[bold #97D6FF]██╔══██╗██╔═══██╗██╔════╝██║██╔══██╗██╔════╝██╔═══██╗████╗  ██║      ██╔══██╗██╔════╝ ██╔════╝████╗  ██║╚══██╔══╝[/]
+[#75C1F6]██████╔╝██║   ██║███████╗██║██║  ██║█████╗  ██║   ██║██╔██╗ ██║█████╗███████║██║  ███╗█████╗  ██╔██╗ ██║   ██║[/]
+[#4FA2E0]██╔═══╝ ██║   ██║╚════██║██║██║  ██║██╔══╝  ██║   ██║██║╚██╗██║╚════╝██╔══██║██║   ██║██╔══╝  ██║╚██╗██║   ██║[/]
+[#2E7CC7]██║     ╚██████╔╝███████║██║██████╔╝███████╗╚██████╔╝██║ ╚████║      ██║  ██║╚██████╔╝███████╗██║ ╚████║   ██║[/]
+[#1B4F95]╚═╝      ╚═════╝ ╚══════╝╚═╝╚═════╝ ╚══════╝ ╚═════╝ ╚═╝  ╚═══╝      ╚═╝  ╚═╝ ╚═════╝ ╚══════╝╚═╝  ╚═══╝   ╚═╝[/]""",
+        "banner_hero": """[#2A6FB9]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#5DB8F5]⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⣾⣿⣷⣄⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#5DB8F5]⠀⠀⠀⠀⠀⠀⠀⢠⣿⠏⠀Ψ⠀⠹⣿⡄⠀⠀⠀⠀⠀⠀⠀[/]
+[#A9DFFF]⠀⠀⠀⠀⠀⠀⠀⣿⡟⠀⠀⠀⠀⠀⢻⣿⠀⠀⠀⠀⠀⠀⠀[/]
+[#A9DFFF]⠀⠀⠀≈≈≈≈≈⣿⡇⠀⠀⠀⠀⠀⢸⣿≈≈≈≈≈⠀⠀⠀[/]
+[#5DB8F5]⠀⠀⠀⠀⠀⠀⠀⣿⡇⠀⠀⠀⠀⠀⢸⣿⠀⠀⠀⠀⠀⠀⠀[/]
+[#2A6FB9]⠀⠀⠀⠀⠀⠀⠀⢿⣧⠀⠀⠀⠀⠀⣼⡿⠀⠀⠀⠀⠀⠀⠀[/]
+[#2A6FB9]⠀⠀⠀⠀⠀⠀⠀⠘⢿⣷⣄⣀⣠⣾⡿⠃⠀⠀⠀⠀⠀⠀⠀[/]
+[#153C73]⠀⠀⠀⠀⠀⠀⠀⠀⠈⠻⣿⣿⡿⠟⠁⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#153C73]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#5DB8F5]⠀⠀⠀⠀⠀≈≈≈≈≈≈≈≈≈≈≈≈≈≈≈⠀⠀⠀⠀⠀[/]
+[#A9DFFF]⠀⠀⠀⠀⠀⠀≈≈≈≈≈≈≈≈≈≈≈≈≈⠀⠀⠀⠀⠀⠀[/]
+[dim #153C73]⠀⠀⠀⠀⠀⠀⠀deep waters hold⠀⠀⠀⠀⠀⠀⠀[/]""",
+    },
+    "sisyphus": {
+        "name": "sisyphus",
+        "description": "Sisyphean theme — austere grayscale with persistence",
+        "colors": {
+            "banner_border": "#B7B7B7",
+            "banner_title": "#F5F5F5",
+            "banner_accent": "#E7E7E7",
+            "banner_dim": "#4A4A4A",
+            "banner_text": "#D3D3D3",
+            "ui_accent": "#E7E7E7",
+            "ui_label": "#D3D3D3",
+            "ui_ok": "#919191",
+            "ui_error": "#E7E7E7",
+            "ui_warn": "#B7B7B7",
+            "prompt": "#F5F5F5",
+            "input_rule": "#656565",
+            "response_border": "#B7B7B7",
+            "session_label": "#919191",
+            "session_border": "#656565",
+        },
+        "spinner": {
+            "waiting_faces": ["(◉)", "(◌)", "(◬)", "(⬤)", "(::)"],
+            "thinking_faces": ["(◉)", "(◬)", "(◌)", "(○)", "(●)"],
+            "thinking_verbs": [
+                "finding traction", "measuring the grade", "resetting the boulder",
+                "counting the ascent", "testing leverage", "setting the shoulder",
+                "pushing uphill", "enduring the loop",
+            ],
+            "wings": [
+                ["⟪◉", "◉⟫"],
+                ["⟪◬", "◬⟫"],
+                ["⟪◌", "◌⟫"],
+                ["⟪⬤", "⬤⟫"],
+            ],
+        },
+        "branding": {
+            "agent_name": "Sisyphus Agent",
+            "welcome": "Welcome to Sisyphus Agent! Type your message or /help for commands.",
+            "goodbye": "The boulder waits. ◉",
+            "response_label": " ◉ Sisyphus ",
+            "prompt_symbol": "◉ ❯ ",
+            "help_header": "(◉) Available Commands",
+        },
+        "tool_prefix": "│",
+        "banner_logo": """[bold #F5F5F5]███████╗██╗███████╗██╗   ██╗██████╗ ██╗  ██╗██╗   ██╗███████╗       █████╗  ██████╗ ███████╗███╗   ██╗████████╗[/]
+[bold #E7E7E7]██╔════╝██║██╔════╝╚██╗ ██╔╝██╔══██╗██║  ██║██║   ██║██╔════╝      ██╔══██╗██╔════╝ ██╔════╝████╗  ██║╚══██╔══╝[/]
+[#D7D7D7]███████╗██║███████╗ ╚████╔╝ ██████╔╝███████║██║   ██║███████╗█████╗███████║██║  ███╗█████╗  ██╔██╗ ██║   ██║[/]
+[#BFBFBF]╚════██║██║╚════██║  ╚██╔╝  ██╔═══╝ ██╔══██║██║   ██║╚════██║╚════╝██╔══██║██║   ██║██╔══╝  ██║╚██╗██║   ██║[/]
+[#8F8F8F]███████║██║███████║   ██║   ██║     ██║  ██║╚██████╔╝███████║      ██║  ██║╚██████╔╝███████╗██║ ╚████║   ██║[/]
+[#626262]╚══════╝╚═╝╚══════╝   ╚═╝   ╚═╝     ╚═╝  ╚═╝ ╚═════╝ ╚══════╝      ╚═╝  ╚═╝ ╚═════╝ ╚══════╝╚═╝  ╚═══╝   ╚═╝[/]""",
+        "banner_hero": """[#B7B7B7]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣀⣀⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#D3D3D3]⠀⠀⠀⠀⠀⠀⠀⣠⣾⣿⣿⣿⣿⣷⣄⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#E7E7E7]⠀⠀⠀⠀⠀⠀⣾⣿⣿⣿⣿⣿⣿⣿⣷⠀⠀⠀⠀⠀⠀⠀[/]
+[#F5F5F5]⠀⠀⠀⠀⠀⢸⣿⣿⣿⣿⣿⣿⣿⣿⣿⡇⠀⠀⠀⠀⠀⠀[/]
+[#E7E7E7]⠀⠀⠀⠀⠀⠀⣿⣿⣿⣿⣿⣿⣿⣿⣿⠀⠀⠀⠀⠀⠀⠀[/]
+[#D3D3D3]⠀⠀⠀⠀⠀⠀⠘⢿⣿⣿⣿⣿⣿⡿⠃⠀⠀⠀⠀⠀⠀⠀[/]
+[#B7B7B7]⠀⠀⠀⠀⠀⠀⠀⠀⠙⠿⣿⠿⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#919191]⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#656565]⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⡄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#656565]⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#4A4A4A]⠀⠀⠀⠀⠀⠀⠀⣰⣿⣿⣿⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#4A4A4A]⠀⠀⠀⠀⠀⣀⣴⣿⣿⣿⣿⣿⣿⣦⣀⠀⠀⠀⠀⠀⠀[/]
+[#656565]⠀⠀⠀━━━━━━━━━━━━━━━━━━━━━━━⠀⠀⠀[/]
+[dim #4A4A4A]⠀⠀⠀⠀⠀⠀⠀⠀⠀the boulder⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]""",
+    },
+    "charizard": {
+        "name": "charizard",
+        "description": "Volcanic theme — burnt orange and ember",
+        "colors": {
+            "banner_border": "#C75B1D",
+            "banner_title": "#FFD39A",
+            "banner_accent": "#F29C38",
+            "banner_dim": "#7A3511",
+            "banner_text": "#FFF0D4",
+            "ui_accent": "#F29C38",
+            "ui_label": "#FFD39A",
+            "ui_ok": "#4caf50",
+            "ui_error": "#ef5350",
+            "ui_warn": "#ffa726",
+            "prompt": "#FFF0D4",
+            "input_rule": "#C75B1D",
+            "response_border": "#F29C38",
+            "session_label": "#FFD39A",
+            "session_border": "#6C4724",
+        },
+        "spinner": {
+            "waiting_faces": ["(✦)", "(▲)", "(◇)", "(<>)", "(🔥)"],
+            "thinking_faces": ["(✦)", "(▲)", "(◇)", "(⌁)", "(🔥)"],
+            "thinking_verbs": [
+                "banking into the draft", "measuring burn", "reading the updraft",
+                "tracking ember fall", "setting wing angle", "holding the flame core",
+                "plotting a hot landing", "coiling for lift",
+            ],
+            "wings": [
+                ["⟪✦", "✦⟫"],
+                ["⟪▲", "▲⟫"],
+                ["⟪◌", "◌⟫"],
+                ["⟪◇", "◇⟫"],
+            ],
+        },
+        "branding": {
+            "agent_name": "Charizard Agent",
+            "welcome": "Welcome to Charizard Agent! Type your message or /help for commands.",
+            "goodbye": "Flame out! ✦",
+            "response_label": " ✦ Charizard ",
+            "prompt_symbol": "✦ ❯ ",
+            "help_header": "(✦) Available Commands",
+        },
+        "tool_prefix": "│",
+        "banner_logo": """[bold #FFF0D4] ██████╗██╗  ██╗ █████╗ ██████╗ ██╗███████╗ █████╗ ██████╗ ██████╗        █████╗  ██████╗ ███████╗███╗   ██╗████████╗[/]
+[bold #FFD39A]██╔════╝██║  ██║██╔══██╗██╔══██╗██║╚══███╔╝██╔══██╗██╔══██╗██╔══██╗      ██╔══██╗██╔════╝ ██╔════╝████╗  ██║╚══██╔══╝[/]
+[#F29C38]██║     ███████║███████║██████╔╝██║  ███╔╝ ███████║██████╔╝██║  ██║█████╗███████║██║  ███╗█████╗  ██╔██╗ ██║   ██║[/]
+[#E2832B]██║     ██╔══██║██╔══██║██╔══██╗██║ ███╔╝  ██╔══██║██╔══██╗██║  ██║╚════╝██╔══██║██║   ██║██╔══╝  ██║╚██╗██║   ██║[/]
+[#C75B1D]╚██████╗██║  ██║██║  ██║██║  ██║██║███████╗██║  ██║██║  ██║██████╔╝      ██║  ██║╚██████╔╝███████╗██║ ╚████║   ██║[/]
+[#7A3511] ╚═════╝╚═╝  ╚═╝╚═╝  ╚═╝╚═╝  ╚═╝╚═╝╚══════╝╚═╝  ╚═╝╚═╝  ╚═╝╚═════╝       ╚═╝  ╚═╝ ╚═════╝ ╚══════╝╚═╝  ╚═══╝   ╚═╝[/]""",
+        "banner_hero": """[#FFD39A]⠀⠀⠀⠀⠀⠀⠀⠀⣀⣤⠶⠶⠶⣤⣀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#F29C38]⠀⠀⠀⠀⠀⠀⣴⠟⠁⠀⠀⠀⠀⠈⠻⣦⠀⠀⠀⠀⠀⠀[/]
+[#F29C38]⠀⠀⠀⠀⠀⣼⠏⠀⠀⠀✦⠀⠀⠀⠀⠹⣧⠀⠀⠀⠀⠀[/]
+[#E2832B]⠀⠀⠀⠀⢰⡟⠀⠀⣀⣤⣤⣤⣀⠀⠀⠀⢻⡆⠀⠀⠀⠀[/]
+[#E2832B]⠀⠀⣠⡾⠛⠁⣠⣾⠟⠉⠀⠉⠻⣷⣄⠀⠈⠛⢷⣄⠀⠀[/]
+[#C75B1D]⠀⣼⠟⠀⢀⣾⠟⠁⠀⠀⠀⠀⠀⠈⠻⣷⡀⠀⠻⣧⠀[/]
+[#C75B1D]⢸⡟⠀⠀⣿⡟⠀⠀⠀🔥⠀⠀⠀⠀⢻⣿⠀⠀⢻⡇[/]
+[#7A3511]⠀⠻⣦⡀⠘⢿⣧⡀⠀⠀⠀⠀⠀⢀⣼⡿⠃⢀⣴⠟⠀[/]
+[#7A3511]⠀⠀⠈⠻⣦⣀⠙⢿⣷⣤⣤⣤⣾⡿⠋⣀⣴⠟⠁⠀⠀[/]
+[#C75B1D]⠀⠀⠀⠀⠈⠙⠛⠶⠤⠭⠭⠤⠶⠛⠋⠁⠀⠀⠀⠀[/]
+[#F29C38]⠀⠀⠀⠀⠀⠀⠀⠀⣰⡿⢿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[#F29C38]⠀⠀⠀⠀⠀⠀⠀⣼⡟⠀⠀⢻⣧⠀⠀⠀⠀⠀⠀⠀⠀[/]
+[dim #7A3511]⠀⠀⠀⠀⠀⠀⠀tail flame lit⠀⠀⠀⠀⠀⠀⠀⠀[/]""",
+    },
+}
+
+
+# =============================================================================
+# Skin loading and management
+# =============================================================================
+
+_active_skin: Optional[SkinConfig] = None
+_active_skin_name: str = "default"
+
+
+def _skins_dir() -> Path:
+    """User skins directory."""
+    home = Path(os.getenv("HERMES_HOME", Path.home() / ".hermes"))
+    return home / "skins"
+
+
+def _load_skin_from_yaml(path: Path) -> Optional[Dict[str, Any]]:
+    """Load a skin definition from a YAML file."""
+    try:
+        import yaml
+        with open(path, "r", encoding="utf-8") as f:
+            data = yaml.safe_load(f)
+        if isinstance(data, dict) and "name" in data:
+            return data
+    except Exception as e:
+        logger.debug("Failed to load skin from %s: %s", path, e)
+    return None
+
+
+def _build_skin_config(data: Dict[str, Any]) -> SkinConfig:
+    """Build a SkinConfig from a raw dict (built-in or loaded from YAML)."""
+    # Start with default values as base for missing keys
+    default = _BUILTIN_SKINS["default"]
+    colors = dict(default.get("colors", {}))
+    colors.update(data.get("colors", {}))
+    spinner = dict(default.get("spinner", {}))
+    spinner.update(data.get("spinner", {}))
+    branding = dict(default.get("branding", {}))
+    branding.update(data.get("branding", {}))
+
+    return SkinConfig(
+        name=data.get("name", "unknown"),
+        description=data.get("description", ""),
+        colors=colors,
+        spinner=spinner,
+        branding=branding,
+        tool_prefix=data.get("tool_prefix", default.get("tool_prefix", "┊")),
+        banner_logo=data.get("banner_logo", ""),
+        banner_hero=data.get("banner_hero", ""),
+    )
+
+
+def list_skins() -> List[Dict[str, str]]:
+    """List all available skins (built-in + user-installed).
+
+    Returns list of {"name": ..., "description": ..., "source": "builtin"|"user"}.
+    """
+    result = []
+    for name, data in _BUILTIN_SKINS.items():
+        result.append({
+            "name": name,
+            "description": data.get("description", ""),
+            "source": "builtin",
+        })
+
+    skins_path = _skins_dir()
+    if skins_path.is_dir():
+        for f in sorted(skins_path.glob("*.yaml")):
+            data = _load_skin_from_yaml(f)
+            if data:
+                skin_name = data.get("name", f.stem)
+                # Skip if it shadows a built-in
+                if any(s["name"] == skin_name for s in result):
+                    continue
+                result.append({
+                    "name": skin_name,
+                    "description": data.get("description", ""),
+                    "source": "user",
+                })
+
+    return result
+
+
+def load_skin(name: str) -> SkinConfig:
+    """Load a skin by name. Checks user skins first, then built-in."""
+    # Check user skins directory
+    skins_path = _skins_dir()
+    user_file = skins_path / f"{name}.yaml"
+    if user_file.is_file():
+        data = _load_skin_from_yaml(user_file)
+        if data:
+            return _build_skin_config(data)
+
+    # Check built-in skins
+    if name in _BUILTIN_SKINS:
+        return _build_skin_config(_BUILTIN_SKINS[name])
+
+    # Fallback to default
+    logger.warning("Skin '%s' not found, using default", name)
+    return _build_skin_config(_BUILTIN_SKINS["default"])
+
+
+def get_active_skin() -> SkinConfig:
+    """Get the currently active skin config (cached)."""
+    global _active_skin
+    if _active_skin is None:
+        _active_skin = load_skin(_active_skin_name)
+    return _active_skin
+
+
+def set_active_skin(name: str) -> SkinConfig:
+    """Switch the active skin. Returns the new SkinConfig."""
+    global _active_skin, _active_skin_name
+    _active_skin_name = name
+    _active_skin = load_skin(name)
+    return _active_skin
+
+
+def get_active_skin_name() -> str:
+    """Get the name of the currently active skin."""
+    return _active_skin_name
+
+
+def init_skin_from_config(config: dict) -> None:
+    """Initialize the active skin from CLI config at startup.
+
+    Call this once during CLI init with the loaded config dict.
+    """
+    display = config.get("display", {})
+    skin_name = display.get("skin", "default")
+    if isinstance(skin_name, str) and skin_name.strip():
+        set_active_skin(skin_name.strip())
+    else:
+        set_active_skin("default")
--- a/hermes_cli/status.py
+++ b/hermes_cli/status.py
@@ -206,6 +206,8 @@ def show_status(args):
        "Telegram": ("TELEGRAM_BOT_TOKEN", "TELEGRAM_HOME_CHANNEL"),
        "Discord": ("DISCORD_BOT_TOKEN", "DISCORD_HOME_CHANNEL"),
        "WhatsApp": ("WHATSAPP_ENABLED", None),
+        "Signal": ("SIGNAL_HTTP_URL", "SIGNAL_HOME_CHANNEL"),
+        "Slack": ("SLACK_BOT_TOKEN", None),
    }
    
    for name, (token_var, home_var) in platforms.items():
@@ -261,7 +263,7 @@ def show_status(args):
    if jobs_file.exists():
        import json
        try:
-            with open(jobs_file) as f:
+            with open(jobs_file, encoding="utf-8") as f:
                data = json.load(f)
                jobs = data.get("jobs", [])
                enabled_jobs = [j for j in jobs if j.get("enabled", True)]
@@ -281,7 +283,7 @@ def show_status(args):
    if sessions_file.exists():
        import json
        try:
-            with open(sessions_file) as f:
+            with open(sessions_file, encoding="utf-8") as f:
                data = json.load(f)
                print(f"  Active:       {len(data)} session(s)")
        except Exception:
--- a/hermes_cli/tools_config.py
+++ b/hermes_cli/tools_config.py
@@ -96,6 +96,11 @@ CONFIGURABLE_TOOLSETS = [
    ("homeassistant",    "🏠 Home Assistant",           "smart home device control"),
 ]

+# Toolsets that are OFF by default for new installs.
+# They're still in _HERMES_CORE_TOOLS (available at runtime if enabled),
+# but the setup checklist won't pre-select them for first-time users.
+_DEFAULT_OFF_TOOLSETS = {"moa", "homeassistant", "rl"}
+
 # Platform display config
 PLATFORMS = {
    "cli":      {"label": "🖥️  CLI",       "default_toolset": "hermes-cli"},
@@ -142,6 +147,8 @@ TOOL_CATEGORIES = {
    },
    "web": {
        "name": "Web Search & Extract",
+        "setup_title": "Select Search Provider",
+        "setup_note": "A free DuckDuckGo search skill is also included — skip this if you don't need Firecrawl.",
        "icon": "🔍",
        "providers": [
            {
@@ -595,11 +602,18 @@ def _configure_tool_category(ts_key: str, cat: dict, config: dict):
        print(color(f"  --- {icon} {name} ({provider['name']}) ---", Colors.CYAN))
        if provider.get("tag"):
            _print_info(f"  {provider['tag']}")
+        # For single-provider tools, show a note if available
+        if cat.get("setup_note"):
+            _print_info(f"  {cat['setup_note']}")
        _configure_provider(provider, config)
    else:
        # Multiple providers - let user choose
        print()
-        print(color(f"  --- {icon} {name} - Choose a provider ---", Colors.CYAN))
+        # Use custom title if provided (e.g. "Select Search Provider")
+        title = cat.get("setup_title", f"Choose a provider")
+        print(color(f"  --- {icon} {name} - {title} ---", Colors.CYAN))
+        if cat.get("setup_note"):
+            _print_info(f"  {cat['setup_note']}")
        print()

        # Plain text labels only (no ANSI codes in menu items)
@@ -617,6 +631,9 @@ def _configure_tool_category(ts_key: str, cat: dict, config: dict):
                    configured = " [configured]"
            provider_choices.append(f"{p['name']}{tag}{configured}")

+        # Add skip option
+        provider_choices.append("Skip — keep defaults / configure later")
+
        # Detect current provider as default
        default_idx = 0
        for i, p in enumerate(providers):
@@ -628,7 +645,13 @@ def _configure_tool_category(ts_key: str, cat: dict, config: dict):
                default_idx = i
                break

-        provider_idx = _prompt_choice("  Select provider:", provider_choices, default_idx)
+        provider_idx = _prompt_choice(f"  {title}:", provider_choices, default_idx)
+
+        # Skip selected
+        if provider_idx >= len(providers):
+            _print_info(f"  Skipped {name}")
+            return
+
        _configure_provider(providers[provider_idx], config)


@@ -835,9 +858,19 @@ def _reconfigure_simple_requirements(ts_key: str):

 # ─── Main Entry Point ─────────────────────────────────────────────────────────

-def tools_command(args=None):
-    """Entry point for `hermes tools` and `hermes setup tools`."""
-    config = load_config()
+def tools_command(args=None, first_install: bool = False, config: dict = None):
+    """Entry point for `hermes tools` and `hermes setup tools`.
+
+    Args:
+        first_install: When True (set by the setup wizard on fresh installs),
+            skip the platform menu, go straight to the CLI checklist, and
+            prompt for API keys on all enabled tools that need them.
+        config: Optional config dict to use.  When called from the setup
+            wizard, the wizard passes its own dict so that platform_toolsets
+            are written into it and survive the wizard's final save_config().
+    """
+    if config is None:
+        config = load_config()
    enabled_platforms = _get_enabled_platforms()

    print()
@@ -846,6 +879,57 @@ def tools_command(args=None):
    print(color("  Tools that need API keys will be configured when enabled.", Colors.DIM))
    print()

+    # ── First-time install: linear flow, no platform menu ──
+    if first_install:
+        for pkey in enabled_platforms:
+            pinfo = PLATFORMS[pkey]
+            current_enabled = _get_platform_tools(config, pkey)
+
+            # Uncheck toolsets that should be off by default
+            checklist_preselected = current_enabled - _DEFAULT_OFF_TOOLSETS
+
+            # Show checklist
+            new_enabled = _prompt_toolset_checklist(pinfo["label"], checklist_preselected)
+
+            added = new_enabled - current_enabled
+            removed = current_enabled - new_enabled
+            if added:
+                for ts in sorted(added):
+                    label = next((l for k, l, _ in CONFIGURABLE_TOOLSETS if k == ts), ts)
+                    print(color(f"  + {label}", Colors.GREEN))
+            if removed:
+                for ts in sorted(removed):
+                    label = next((l for k, l, _ in CONFIGURABLE_TOOLSETS if k == ts), ts)
+                    print(color(f"  - {label}", Colors.RED))
+
+            # Walk through ALL selected tools that have provider options or
+            # need API keys.  This ensures browser (Local vs Browserbase),
+            # TTS (Edge vs OpenAI vs ElevenLabs), etc. are shown even when
+            # a free provider exists.
+            to_configure = [
+                ts_key for ts_key in sorted(new_enabled)
+                if TOOL_CATEGORIES.get(ts_key) or TOOLSET_ENV_REQUIREMENTS.get(ts_key)
+            ]
+
+            if to_configure:
+                print()
+                print(color(f"  Configuring {len(to_configure)} tool(s):", Colors.YELLOW))
+                for ts_key in to_configure:
+                    label = next((l for k, l, _ in CONFIGURABLE_TOOLSETS if k == ts_key), ts_key)
+                    print(color(f"    • {label}", Colors.DIM))
+                print(color("  You can skip any tool you don't need right now.", Colors.DIM))
+                print()
+                for ts_key in to_configure:
+                    _configure_toolset(ts_key, config)
+
+            _save_platform_tools(config, pkey, new_enabled)
+            save_config(config)
+            print(color(f"  ✓ Saved {pinfo['label']} tool configuration", Colors.GREEN))
+            print()
+
+        return
+
+    # ── Returning user: platform menu loop ──
    # Build platform choices
    platform_choices = []
    platform_keys = []
@@ -896,11 +980,10 @@ def tools_command(args=None):
                    print(color(f"  - {label}", Colors.RED))

            # Configure newly enabled toolsets that need API keys
-            if added:
-                for ts_key in sorted(added):
-                    if TOOL_CATEGORIES.get(ts_key) or TOOLSET_ENV_REQUIREMENTS.get(ts_key):
-                        if not _toolset_has_keys(ts_key):
-                            _configure_toolset(ts_key, config)
+            for ts_key in sorted(added):
+                if (TOOL_CATEGORIES.get(ts_key) or TOOLSET_ENV_REQUIREMENTS.get(ts_key)):
+                    if not _toolset_has_keys(ts_key):
+                        _configure_toolset(ts_key, config)

            _save_platform_tools(config, pkey, new_enabled)
            save_config(config)
--- a/hermes_constants.py
+++ b/hermes_constants.py
@@ -7,3 +7,6 @@ without risk of circular imports.
 OPENROUTER_BASE_URL = "https://openrouter.ai/api/v1"
 OPENROUTER_MODELS_URL = f"{OPENROUTER_BASE_URL}/models"
 OPENROUTER_CHAT_URL = f"{OPENROUTER_BASE_URL}/chat/completions"
+
+NOUS_API_BASE_URL = "https://inference-api.nousresearch.com/v1"
+NOUS_API_CHAT_URL = f"{NOUS_API_BASE_URL}/chat/completions"
--- a/hermes_state.py
+++ b/hermes_state.py
@@ -16,6 +16,7 @@ Key design decisions:

 import json
 import os
+import re
 import sqlite3
 import time
 from pathlib import Path
@@ -490,12 +491,16 @@ class SessionDB:
        msg_id = cursor.lastrowid

        # Update counters
-        is_tool_related = role == "tool" or tool_calls is not None
-        if is_tool_related:
+        # Count actual tool calls from the tool_calls list (not from tool responses).
+        # A single assistant message can contain multiple parallel tool calls.
+        num_tool_calls = 0
+        if tool_calls is not None:
+            num_tool_calls = len(tool_calls) if isinstance(tool_calls, list) else 1
+        if num_tool_calls > 0:
            self._conn.execute(
                """UPDATE sessions SET message_count = message_count + 1,
-                   tool_call_count = tool_call_count + 1 WHERE id = ?""",
-                (session_id,),
+                   tool_call_count = tool_call_count + ? WHERE id = ?""",
+                (num_tool_calls, session_id),
            )
        else:
            self._conn.execute(
@@ -553,6 +558,32 @@ class SessionDB:
    # Search
    # =========================================================================

+    @staticmethod
+    def _sanitize_fts5_query(query: str) -> str:
+        """Sanitize user input for safe use in FTS5 MATCH queries.
+
+        FTS5 has its own query syntax where characters like ``"``, ``(``, ``)``,
+        ``+``, ``*``, ``{``, ``}`` and bare boolean operators (``AND``, ``OR``,
+        ``NOT``) have special meaning.  Passing raw user input directly to
+        MATCH can cause ``sqlite3.OperationalError``.
+
+        Strategy: strip characters that are only meaningful as FTS5 operators
+        and would otherwise cause syntax errors.  This preserves normal keyword
+        search while preventing crashes on inputs like ``C++``, ``"unterminated``,
+        or ``hello AND``.
+        """
+        # Remove FTS5-special characters that are not useful in keyword search
+        sanitized = re.sub(r'[+{}()"^]', " ", query)
+        # Collapse repeated * (e.g. "***") into a single one, and remove
+        # leading * (prefix-only matching requires at least one char before *)
+        sanitized = re.sub(r"\*+", "*", sanitized)
+        sanitized = re.sub(r"(^|\s)\*", r"\1", sanitized)
+        # Remove dangling boolean operators at start/end that would cause
+        # syntax errors (e.g. "hello AND" or "OR world")
+        sanitized = re.sub(r"(?i)^(AND|OR|NOT)\b\s*", "", sanitized.strip())
+        sanitized = re.sub(r"(?i)\s+(AND|OR|NOT)\s*$", "", sanitized.strip())
+        return sanitized.strip()
+
    def search_messages(
        self,
        query: str,
@@ -576,6 +607,10 @@ class SessionDB:
        if not query or not query.strip():
            return []

+        query = self._sanitize_fts5_query(query)
+        if not query:
+            return []
+
        if source_filter is None:
            source_filter = ["cli", "telegram", "discord", "whatsapp", "slack"]

@@ -615,7 +650,11 @@ class SessionDB:
            LIMIT ? OFFSET ?
        """

-        cursor = self._conn.execute(sql, params)
+        try:
+            cursor = self._conn.execute(sql, params)
+        except sqlite3.OperationalError:
+            # FTS5 query syntax error despite sanitization — return empty
+            return []
        matches = [dict(row) for row in cursor.fetchall()]

        # Add surrounding context (1 message before + after each match)
--- a/landingpage/apple-touch-icon.png
+++ b/landingpage/apple-touch-icon.png
--- a/landingpage/favicon-16x16.png
+++ b/landingpage/favicon-16x16.png
--- a/landingpage/favicon-32x32.png
+++ b/landingpage/favicon-32x32.png
--- a/landingpage/favicon.ico
+++ b/landingpage/favicon.ico
--- a/landingpage/icon-192.png
+++ b/landingpage/icon-192.png
--- a/landingpage/icon-512.png
+++ b/landingpage/icon-512.png
--- a/landingpage/index.html
+++ b/landingpage/index.html
@@ -19,7 +19,10 @@
    <link href="https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700&family=JetBrains+Mono:wght@400;500&display=swap" rel="stylesheet">
    
    <link rel="stylesheet" href="style.css">
-    <link rel="icon" href="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 100 100'><text y='.9em' font-size='90'>⚕</text></svg>">
+    <link rel="icon" type="image/x-icon" href="favicon.ico">
+    <link rel="icon" type="image/png" sizes="32x32" href="favicon-32x32.png">
+    <link rel="icon" type="image/png" sizes="16x16" href="favicon-16x16.png">
+    <link rel="apple-touch-icon" sizes="180x180" href="apple-touch-icon.png">
 </head>
 <body>
    <!-- Ambient glow effects -->
--- a/model_tools.py
+++ b/model_tools.py
@@ -266,6 +266,7 @@ def handle_function_call(
    function_args: Dict[str, Any],
    task_id: Optional[str] = None,
    user_task: Optional[str] = None,
+    enabled_tools: Optional[List[str]] = None,
 ) -> str:
    """
    Main function call dispatcher that routes calls to the tool registry.
@@ -275,6 +276,10 @@ def handle_function_call(
        function_args: Arguments for the function.
        task_id: Unique identifier for terminal/browser session isolation.
        user_task: The user's original task (for browser_snapshot context).
+        enabled_tools: Tool names enabled for this session.  When provided,
+                       execute_code uses this list to determine which sandbox
+                       tools to generate.  Falls back to the process-global
+                       ``_last_resolved_tool_names`` for backward compat.

    Returns:
        Function result as a JSON string.
@@ -284,10 +289,13 @@ def handle_function_call(
            return json.dumps({"error": f"{function_name} must be handled by the agent loop"})

        if function_name == "execute_code":
+            # Prefer the caller-provided list so subagents can't overwrite
+            # the parent's tool set via the process-global.
+            sandbox_enabled = enabled_tools if enabled_tools is not None else _last_resolved_tool_names
            return registry.dispatch(
                function_name, function_args,
                task_id=task_id,
-                enabled_tools=_last_resolved_tool_names,
+                enabled_tools=sandbox_enabled,
            )

        return registry.dispatch(
--- a/optional-skills/email/agentmail/SKILL.md
+++ b/optional-skills/email/agentmail/SKILL.md
@@ -0,0 +1,125 @@
+---
+name: agentmail
+description: Give the agent its own dedicated email inbox via AgentMail. Send, receive, and manage email autonomously using agent-owned email addresses (e.g. hermes-agent@agentmail.to).
+version: 1.0.0
+metadata:
+  hermes:
+    tags: [email, communication, agentmail, mcp]
+    category: email
+---
+
+# AgentMail — Agent-Owned Email Inboxes
+
+## Requirements
+
+- **AgentMail API key** (required) — sign up at https://console.agentmail.to (free tier: 3 inboxes, 3,000 emails/month; paid plans from $20/mo)
+- Node.js 18+ (for the MCP server)
+
+## When to Use
+Use this skill when you need to:
+- Give the agent its own dedicated email address
+- Send emails autonomously on behalf of the agent
+- Receive and read incoming emails
+- Manage email threads and conversations
+- Sign up for services or authenticate via email
+- Communicate with other agents or humans via email
+
+This is NOT for reading the user's personal email (use himalaya or Gmail for that).
+AgentMail gives the agent its own identity and inbox.
+
+## Setup
+
+### 1. Get an API Key
+- Go to https://console.agentmail.to
+- Create an account and generate an API key (starts with `am_`)
+
+### 2. Configure MCP Server
+Add to `~/.hermes/config.yaml` (paste your actual key — MCP env vars are not expanded from .env):
+```yaml
+mcp_servers:
+  agentmail:
+    command: "npx"
+    args: ["-y", "agentmail-mcp"]
+    env:
+      AGENTMAIL_API_KEY: "am_your_key_here"
+```
+
+### 3. Restart Hermes
+```bash
+hermes
+```
+All 11 AgentMail tools are now available automatically.
+
+## Available Tools (via MCP)
+
+| Tool | Description |
+|------|-------------|
+| `list_inboxes` | List all agent inboxes |
+| `get_inbox` | Get details of a specific inbox |
+| `create_inbox` | Create a new inbox (gets a real email address) |
+| `delete_inbox` | Delete an inbox |
+| `list_threads` | List email threads in an inbox |
+| `get_thread` | Get a specific email thread |
+| `send_message` | Send a new email |
+| `reply_to_message` | Reply to an existing email |
+| `forward_message` | Forward an email |
+| `update_message` | Update message labels/status |
+| `get_attachment` | Download an email attachment |
+
+## Procedure
+
+### Create an inbox and send an email
+1. Create a dedicated inbox:
+   - Use `create_inbox` with a username (e.g. `hermes-agent`)
+   - The agent gets address: `hermes-agent@agentmail.to`
+2. Send an email:
+   - Use `send_message` with `inbox_id`, `to`, `subject`, `text`
+3. Check for replies:
+   - Use `list_threads` to see incoming conversations
+   - Use `get_thread` to read a specific thread
+
+### Check incoming email
+1. Use `list_inboxes` to find your inbox ID
+2. Use `list_threads` with the inbox ID to see conversations
+3. Use `get_thread` to read a thread and its messages
+
+### Reply to an email
+1. Get the thread with `get_thread`
+2. Use `reply_to_message` with the message ID and your reply text
+
+## Example Workflows
+
+**Sign up for a service:**
+```
+1. create_inbox (username: "signup-bot")
+2. Use the inbox address to register on the service
+3. list_threads to check for verification email
+4. get_thread to read the verification code
+```
+
+**Agent-to-human outreach:**
+```
+1. create_inbox (username: "hermes-outreach")
+2. send_message (to: user@example.com, subject: "Hello", text: "...")
+3. list_threads to check for replies
+```
+
+## Pitfalls
+- Free tier limited to 3 inboxes and 3,000 emails/month
+- Emails come from `@agentmail.to` domain on free tier (custom domains on paid plans)
+- Node.js (18+) is required for the MCP server (`npx -y agentmail-mcp`)
+- The `mcp` Python package must be installed: `pip install mcp`
+- Real-time inbound email (webhooks) requires a public server — use `list_threads` polling via cronjob instead for personal use
+
+## Verification
+After setup, test with:
+```
+hermes --toolsets mcp -q "Create an AgentMail inbox called test-agent and tell me its email address"
+```
+You should see the new inbox address returned.
+
+## References
+- AgentMail docs: https://docs.agentmail.to/
+- AgentMail console: https://console.agentmail.to
+- AgentMail MCP repo: https://github.com/agentmail-to/agentmail-mcp
+- Pricing: https://www.agentmail.to/pricing
--- a/optional-skills/migration/DESCRIPTION.md
+++ b/optional-skills/migration/DESCRIPTION.md
@@ -0,0 +1,2 @@
+Optional migration workflows for importing user state and customizations from
+other agent systems into Hermes Agent.
--- a/optional-skills/migration/openclaw-migration/SKILL.md
+++ b/optional-skills/migration/openclaw-migration/SKILL.md
@@ -0,0 +1,281 @@
+---
+name: openclaw-migration
+description: Migrate a user's OpenClaw customization footprint into Hermes Agent. Imports Hermes-compatible memories, SOUL.md, command allowlists, user skills, and selected workspace assets from ~/.openclaw, then reports exactly what could not be migrated and why.
+version: 1.0.0
+author: Hermes Agent (Nous Research)
+license: MIT
+metadata:
+  hermes:
+    tags: [Migration, OpenClaw, Hermes, Memory, Persona, Import]
+    related_skills: [hermes-agent]
+---
+
+# OpenClaw -> Hermes Migration
+
+Use this skill when a user wants to move their OpenClaw setup into Hermes Agent with minimal manual cleanup.
+
+## What this skill does
+
+It uses `scripts/openclaw_to_hermes.py` to:
+
+- import `SOUL.md` into the Hermes home directory as `SOUL.md`
+- transform OpenClaw `MEMORY.md` and `USER.md` into Hermes memory entries
+- merge OpenClaw command approval patterns into Hermes `command_allowlist`
+- migrate Hermes-compatible messaging settings such as `TELEGRAM_ALLOWED_USERS` and `MESSAGING_CWD`
+- copy OpenClaw skills into `~/.hermes/skills/openclaw-imports/`
+- optionally copy the OpenClaw workspace instructions file into a chosen Hermes workspace
+- mirror compatible workspace assets such as `workspace/tts/` into `~/.hermes/tts/`
+- archive non-secret docs that do not have a direct Hermes destination
+- produce a structured report listing migrated items, conflicts, skipped items, and reasons
+
+## Path resolution
+
+The helper script lives in this skill directory at:
+
+- `scripts/openclaw_to_hermes.py`
+
+When this skill is installed from the Skills Hub, the normal location is:
+
+- `~/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py`
+
+Do not guess a shorter path like `~/.hermes/skills/openclaw-migration/...`.
+
+Before running the helper:
+
+1. Prefer the installed path under `~/.hermes/skills/migration/openclaw-migration/`.
+2. If that path fails, inspect the installed skill directory and resolve the script relative to the installed `SKILL.md`.
+3. Only use `find` as a fallback if the installed location is missing or the skill was moved manually.
+4. When calling the terminal tool, do not pass `workdir: "~"`. Use an absolute directory such as the user's home directory, or omit `workdir` entirely.
+
+With `--migrate-secrets`, it will also import a small allowlisted set of Hermes-compatible secrets, currently:
+
+- `TELEGRAM_BOT_TOKEN`
+
+## Default workflow
+
+1. Inspect first with a dry run.
+2. Present a simple summary of what can be migrated, what cannot be migrated, and what would be archived.
+3. If the `clarify` tool is available, use it for user decisions instead of asking for a free-form prose reply.
+4. If the dry run finds imported skill directory conflicts, ask how those should be handled before executing.
+5. Ask the user to choose between the two supported migration modes before executing.
+6. Ask for a target workspace path only if the user wants the workspace instructions file brought over.
+7. Execute the migration with the matching preset and flags.
+8. Summarize the results, especially:
+   - what was migrated
+   - what was archived for manual review
+   - what was skipped and why
+
+## User interaction protocol
+
+Hermes CLI supports the `clarify` tool for interactive prompts, but it is limited to:
+
+- one choice at a time
+- up to 4 predefined choices
+- an automatic `Other` free-text option
+
+It does **not** support true multi-select checkboxes in a single prompt.
+
+For every `clarify` call:
+
+- always include a non-empty `question`
+- include `choices` only for real selectable prompts
+- keep `choices` to 2-4 plain string options
+- never emit placeholder or truncated options such as `...`
+- never pad or stylize choices with extra whitespace
+- never include fake form fields in the question such as `enter directory here`, blank lines to fill in, or underscores like `_____`
+- for open-ended path questions, ask only the plain sentence; the user types in the normal CLI prompt below the panel
+
+If a `clarify` call returns an error, inspect the error text, correct the payload, and retry once with a valid `question` and clean choices.
+
+When `clarify` is available and the dry run reveals any required user decision, your **next action must be a `clarify` tool call**.
+Do not end the turn with a normal assistant message such as:
+
+- "Let me present the choices"
+- "What would you like to do?"
+- "Here are the options"
+
+If a user decision is required, collect it via `clarify` before producing more prose.
+If multiple unresolved decisions remain, do not insert an explanatory assistant message between them. After one `clarify` response is received, your next action should usually be the next required `clarify` call.
+
+Treat `workspace-agents` as an unresolved decision whenever the dry run reports:
+
+- `kind="workspace-agents"`
+- `status="skipped"`
+- reason containing `No workspace target was provided`
+
+In that case, you must ask about workspace instructions before execution. Do not silently treat that as a decision to skip.
+
+Because of that limitation, use this simplified decision flow:
+
+1. For `SOUL.md` conflicts, use `clarify` with choices such as:
+   - `keep existing`
+   - `overwrite with backup`
+   - `review first`
+2. If the dry run shows one or more `kind="skill"` items with `status="conflict"`, use `clarify` with choices such as:
+   - `keep existing skills`
+   - `overwrite conflicting skills with backup`
+   - `import conflicting skills under renamed folders`
+3. For workspace instructions, use `clarify` with choices such as:
+   - `skip workspace instructions`
+   - `copy to a workspace path`
+   - `decide later`
+4. If the user chooses to copy workspace instructions, ask a follow-up open-ended `clarify` question requesting an **absolute path**.
+5. If the user chooses `skip workspace instructions` or `decide later`, proceed without `--workspace-target`.
+5. For migration mode, use `clarify` with these 3 choices:
+   - `user-data only`
+   - `full compatible migration`
+   - `cancel`
+6. `user-data only` means: migrate user data and compatible config, but do **not** import allowlisted secrets.
+7. `full compatible migration` means: migrate the same compatible user data plus the allowlisted secrets when present.
+8. If `clarify` is not available, ask the same question in normal text, but still constrain the answer to `user-data only`, `full compatible migration`, or `cancel`.
+
+Execution gate:
+
+- Do not execute while a `workspace-agents` skip caused by `No workspace target was provided` remains unresolved.
+- The only valid ways to resolve it are:
+  - user explicitly chooses `skip workspace instructions`
+  - user explicitly chooses `decide later`
+  - user provides a workspace path after choosing `copy to a workspace path`
+- Absence of a workspace target in the dry run is not itself permission to execute.
+- Do not execute while any required `clarify` decision remains unresolved.
+
+Use these exact `clarify` payload shapes as the default pattern:
+
+- `{"question":"Your existing SOUL.md conflicts with the imported one. What should I do?","choices":["keep existing","overwrite with backup","review first"]}`
+- `{"question":"One or more imported OpenClaw skills already exist in Hermes. How should I handle those skill conflicts?","choices":["keep existing skills","overwrite conflicting skills with backup","import conflicting skills under renamed folders"]}`
+- `{"question":"Choose migration mode: migrate only user data, or run the full compatible migration including allowlisted secrets?","choices":["user-data only","full compatible migration","cancel"]}`
+- `{"question":"Do you want to copy the OpenClaw workspace instructions file into a Hermes workspace?","choices":["skip workspace instructions","copy to a workspace path","decide later"]}`
+- `{"question":"Please provide an absolute path where the workspace instructions should be copied."}`
+
+## Decision-to-command mapping
+
+Map user decisions to command flags exactly:
+
+- If the user chooses `keep existing` for `SOUL.md`, do **not** add `--overwrite`.
+- If the user chooses `overwrite with backup`, add `--overwrite`.
+- If the user chooses `review first`, stop before execution and review the relevant files.
+- If the user chooses `keep existing skills`, add `--skill-conflict skip`.
+- If the user chooses `overwrite conflicting skills with backup`, add `--skill-conflict overwrite`.
+- If the user chooses `import conflicting skills under renamed folders`, add `--skill-conflict rename`.
+- If the user chooses `user-data only`, execute with `--preset user-data` and do **not** add `--migrate-secrets`.
+- If the user chooses `full compatible migration`, execute with `--preset full --migrate-secrets`.
+- Only add `--workspace-target` if the user explicitly provided an absolute workspace path.
+- If the user chooses `skip workspace instructions` or `decide later`, do not add `--workspace-target`.
+
+Before executing, restate the exact command plan in plain language and make sure it matches the user's choices.
+
+## Post-run reporting rules
+
+After execution, treat the script's JSON output as the source of truth.
+
+1. Base all counts on `report.summary`.
+2. Only list an item under "Successfully Migrated" if its `status` is exactly `migrated`.
+3. Do not claim a conflict was resolved unless the report shows that item as `migrated`.
+4. Do not say `SOUL.md` was overwritten unless the report item for `kind="soul"` has `status="migrated"`.
+5. If `report.summary.conflict > 0`, include a conflict section instead of silently implying success.
+6. If counts and listed items disagree, fix the list to match the report before responding.
+7. Include the `output_dir` path from the report when available so the user can inspect `report.json`, `summary.md`, backups, and archived files.
+8. For memory or user-profile overflow, do not say the entries were archived unless the report explicitly shows an archive path. If `details.overflow_file` exists, say the full overflow list was exported there.
+9. If a skill was imported under a renamed folder, report the final destination and mention `details.renamed_from`.
+10. If `report.skill_conflict_mode` is present, use it as the source of truth for the selected imported-skill conflict policy.
+11. If an item has `status="skipped"`, do not describe it as overwritten, backed up, migrated, or resolved.
+12. If `kind="soul"` has `status="skipped"` with reason `Target already matches source`, say it was left unchanged and do not mention a backup.
+13. If a renamed imported skill has an empty `details.backup`, do not imply the existing Hermes skill was renamed or backed up. Say only that the imported copy was placed in the new destination and reference `details.renamed_from` as the pre-existing folder that remained in place.
+
+## Migration presets
+
+Prefer these two presets in normal use:
+
+- `user-data`
+- `full`
+
+`user-data` includes:
+
+- `soul`
+- `workspace-agents`
+- `memory`
+- `user-profile`
+- `messaging-settings`
+- `command-allowlist`
+- `skills`
+- `tts-assets`
+- `archive`
+
+`full` includes everything in `user-data` plus:
+
+- `secret-settings`
+
+The helper script still supports category-level `--include` / `--exclude`, but treat that as an advanced fallback rather than the default UX.
+
+## Commands
+
+Dry run with full discovery:
+
+```bash
+python3 ~/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py
+```
+
+When using the terminal tool, prefer an absolute invocation pattern such as:
+
+```json
+{"command":"python3 /home/USER/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py","workdir":"/home/USER"}
+```
+
+Dry run with the user-data preset:
+
+```bash
+python3 ~/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py --preset user-data
+```
+
+Execute a user-data migration:
+
+```bash
+python3 ~/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py --execute --preset user-data --skill-conflict skip
+```
+
+Execute a full compatible migration:
+
+```bash
+python3 ~/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py --execute --preset full --migrate-secrets --skill-conflict skip
+```
+
+Execute with workspace instructions included:
+
+```bash
+python3 ~/.hermes/skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py --execute --preset user-data --skill-conflict rename --workspace-target "/absolute/workspace/path"
+```
+
+Do not use `$PWD` or the home directory as the workspace target by default. Ask for an explicit workspace path first.
+
+## Important rules
+
+1. Run a dry run before writing unless the user explicitly says to proceed immediately.
+2. Do not migrate secrets by default. Tokens, auth blobs, device credentials, and raw gateway config should stay out of Hermes unless the user explicitly asks for secret migration.
+3. Do not silently overwrite non-empty Hermes targets unless the user explicitly wants that. The helper script will preserve backups when overwriting is enabled.
+4. Always give the user the skipped-items report. That report is part of the migration, not an optional extra.
+5. Prefer the primary OpenClaw workspace (`~/.openclaw/workspace/`) over `workspace.default/`. Only use the default workspace as fallback when the primary files are missing.
+6. Even in secret-migration mode, only migrate secrets with a clean Hermes destination. Unsupported auth blobs must still be reported as skipped.
+7. If the dry run shows a large asset copy, a conflicting `SOUL.md`, or overflowed memory entries, call those out separately before execution.
+8. Default to `user-data only` if the user is unsure.
+9. Only include `workspace-agents` when the user has explicitly provided a destination workspace path.
+10. Treat category-level `--include` / `--exclude` as an advanced escape hatch, not the normal flow.
+11. Do not end the dry-run summary with a vague “What would you like to do?” if `clarify` is available. Use structured follow-up prompts instead.
+12. Do not use an open-ended `clarify` prompt when a real choice prompt would work. Prefer selectable choices first, then free text only for absolute paths or file review requests.
+13. After a dry run, never stop after summarizing if there is still an unresolved decision. Use `clarify` immediately for the highest-priority blocking decision.
+14. Priority order for follow-up questions:
+    - `SOUL.md` conflict
+    - imported skill conflicts
+    - migration mode
+    - workspace instructions destination
+15. Do not promise to present choices later in the same message. Present them by actually calling `clarify`.
+16. After the migration-mode answer, explicitly check whether `workspace-agents` is still unresolved. If it is, your next action must be the workspace-instructions `clarify` call.
+17. After any `clarify` answer, if another required decision remains, do not narrate what was just decided. Ask the next required question immediately.
+
+## Expected result
+
+After a successful run, the user should have:
+
+- Hermes persona state imported
+- Hermes memory files populated with converted OpenClaw knowledge
+- OpenClaw skills available under `~/.hermes/skills/openclaw-imports/`
+- a migration report showing any conflicts, omissions, or unsupported data
--- a/optional-skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py
+++ b/optional-skills/migration/openclaw-migration/scripts/openclaw_to_hermes.py
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -40,13 +40,16 @@ dependencies = [
 [project.optional-dependencies]
 modal = ["swe-rex[modal]>=1.4.0"]
 daytona = ["daytona>=0.148.0"]
-dev = ["pytest", "pytest-asyncio"]
+dev = ["pytest", "pytest-asyncio", "mcp>=1.2.0"]
 messaging = ["python-telegram-bot>=20.0", "discord.py>=2.0", "aiohttp>=3.9.0", "slack-bolt>=1.18.0", "slack-sdk>=3.27.0"]
 cron = ["croniter"]
 slack = ["slack-bolt>=1.18.0", "slack-sdk>=3.27.0"]
 cli = ["simple-term-menu"]
 tts-premium = ["elevenlabs"]
-pty = ["ptyprocess>=0.7.0"]
+pty = [
+  "ptyprocess>=0.7.0; sys_platform != 'win32'",
+  "pywinpty>=2.0.0; sys_platform == 'win32'",
+]
 honcho = ["honcho-ai>=2.0.1"]
 mcp = ["mcp>=1.2.0"]
 homeassistant = ["aiohttp>=3.9.0"]
--- a/run_agent.py
+++ b/run_agent.py
@@ -172,6 +172,7 @@ class AIAgent:
        provider_data_collection: str = None,
        session_id: str = None,
        tool_progress_callback: callable = None,
+        thinking_callback: callable = None,
        clarify_callback: callable = None,
        step_callback: callable = None,
        max_tokens: int = None,
@@ -183,6 +184,9 @@ class AIAgent:
        session_db=None,
        honcho_session_key: str = None,
        iteration_budget: "IterationBudget" = None,
+        fallback_model: Dict[str, Any] = None,
+        checkpoints_enabled: bool = False,
+        checkpoint_max_snapshots: int = 50,
    ):
        """
        Initialize the AI Agent.
@@ -255,6 +259,7 @@ class AIAgent:
            self.api_mode = "chat_completions"

        self.tool_progress_callback = tool_progress_callback
+        self.thinking_callback = thinking_callback
        self.clarify_callback = clarify_callback
        self.step_callback = step_callback
        self._last_reported_tool = None  # Track for "new tool" mode
@@ -406,6 +411,17 @@ class AIAgent:
        except Exception as e:
            raise RuntimeError(f"Failed to initialize OpenAI client: {e}")
        
+        # Provider fallback — a single backup model/provider tried when the
+        # primary is exhausted (rate-limit, overload, connection failure).
+        # Config shape: {"provider": "openrouter", "model": "anthropic/claude-sonnet-4"}
+        self._fallback_model = fallback_model if isinstance(fallback_model, dict) else None
+        self._fallback_activated = False
+        if self._fallback_model:
+            fb_p = self._fallback_model.get("provider", "")
+            fb_m = self._fallback_model.get("model", "")
+            if fb_p and fb_m and not self.quiet_mode:
+                print(f"🔄 Fallback model: {fb_m} ({fb_p})")
+
        # Get available tools with filtering
        self.tools = get_tool_definitions(
            enabled_toolsets=enabled_toolsets,
@@ -472,8 +488,16 @@ class AIAgent:
        # Cached system prompt -- built once per session, only rebuilt on compression
        self._cached_system_prompt: Optional[str] = None
        
+        # Filesystem checkpoint manager (transparent — not a tool)
+        from tools.checkpoint_manager import CheckpointManager
+        self._checkpoint_mgr = CheckpointManager(
+            enabled=checkpoints_enabled,
+            max_snapshots=checkpoint_max_snapshots,
+        )
+        
        # SQLite session store (optional -- provided by CLI or gateway)
        self._session_db = session_db
+        self._last_flushed_db_idx = 0  # tracks DB-write cursor to prevent duplicate writes
        if self._session_db:
            try:
                self._session_db.create_session(
@@ -779,45 +803,19 @@ class AIAgent:
        self._save_session_log(messages)
        self._flush_messages_to_session_db(messages, conversation_history)

-    def _log_msg_to_db(self, msg: Dict):
-        """Log a single message to SQLite immediately. Called after each messages.append()."""
-        if not self._session_db:
-            return
-        try:
-            role = msg.get("role", "unknown")
-            content = msg.get("content")
-            tool_calls_data = None
-            if hasattr(msg, "tool_calls") and msg.tool_calls:
-                tool_calls_data = [
-                    {"name": tc.function.name, "arguments": tc.function.arguments}
-                    for tc in msg.tool_calls
-                ]
-            elif isinstance(msg.get("tool_calls"), list):
-                tool_calls_data = msg["tool_calls"]
-            self._session_db.append_message(
-                session_id=self.session_id,
-                role=role,
-                content=content,
-                tool_name=msg.get("tool_name"),
-                tool_calls=tool_calls_data,
-                tool_call_id=msg.get("tool_call_id"),
-                finish_reason=msg.get("finish_reason"),
-            )
-        except Exception as e:
-            logger.debug("Session DB log_msg failed: %s", e)
-
    def _flush_messages_to_session_db(self, messages: List[Dict], conversation_history: List[Dict] = None):
-        """Persist any un-logged messages to the SQLite session store.
+        """Persist any un-flushed messages to the SQLite session store.

-        Called both at the normal end of run_conversation and from every early-
-        return path so that tool calls, tool responses, and assistant messages
-        are never lost even when the conversation errors out.
+        Uses _last_flushed_db_idx to track which messages have already been
+        written, so repeated calls (from multiple exit paths) only write
+        truly new messages — preventing the duplicate-write bug (#860).
        """
        if not self._session_db:
            return
        try:
            start_idx = len(conversation_history) if conversation_history else 0
-            for msg in messages[start_idx:]:
+            flush_from = max(start_idx, self._last_flushed_db_idx)
+            for msg in messages[flush_from:]:
                role = msg.get("role", "unknown")
                content = msg.get("content")
                tool_calls_data = None
@@ -837,6 +835,7 @@ class AIAgent:
                    tool_call_id=msg.get("tool_call_id"),
                    finish_reason=msg.get("finish_reason"),
                )
+            self._last_flushed_db_idx = len(messages)
        except Exception as e:
            logger.debug("Session DB append_message failed: %s", e)

@@ -1419,6 +1418,34 @@ class AIAgent:

        return "\n\n".join(prompt_parts)
    
+    def _repair_tool_call(self, tool_name: str) -> str | None:
+        """Attempt to repair a mismatched tool name before aborting.
+
+        1. Try lowercase
+        2. Try normalized (lowercase + hyphens/spaces -> underscores)
+        3. Try fuzzy match (difflib, cutoff=0.7)
+
+        Returns the repaired name if found in valid_tool_names, else None.
+        """
+        from difflib import get_close_matches
+
+        # 1. Lowercase
+        lowered = tool_name.lower()
+        if lowered in self.valid_tool_names:
+            return lowered
+
+        # 2. Normalize
+        normalized = lowered.replace("-", "_").replace(" ", "_")
+        if normalized in self.valid_tool_names:
+            return normalized
+
+        # 3. Fuzzy match
+        matches = get_close_matches(lowered, self.valid_tool_names, n=1, cutoff=0.7)
+        if matches:
+            return matches[0]
+
+        return None
+
    def _invalidate_system_prompt(self):
        """
        Invalidate the cached system prompt, forcing a rebuild on the next turn.
@@ -2146,6 +2173,141 @@ class AIAgent:
            raise result["error"]
        return result["response"]

+    # ── Provider fallback ──────────────────────────────────────────────────
+
+    # API-key providers: provider → (base_url, [env_var_names])
+    _FALLBACK_API_KEY_PROVIDERS = {
+        "openrouter": (OPENROUTER_BASE_URL, ["OPENROUTER_API_KEY"]),
+        "zai": ("https://api.z.ai/api/paas/v4", ["ZAI_API_KEY", "Z_AI_API_KEY"]),
+        "kimi-coding": ("https://api.moonshot.ai/v1", ["KIMI_API_KEY"]),
+        "minimax": ("https://api.minimax.io/v1", ["MINIMAX_API_KEY"]),
+        "minimax-cn": ("https://api.minimaxi.com/v1", ["MINIMAX_CN_API_KEY"]),
+    }
+
+    # OAuth providers: provider → (resolver_import_path, api_mode)
+    # Each resolver returns {"api_key": ..., "base_url": ...}.
+    _FALLBACK_OAUTH_PROVIDERS = {
+        "openai-codex": ("resolve_codex_runtime_credentials", "codex_responses"),
+        "nous": ("resolve_nous_runtime_credentials", "chat_completions"),
+    }
+
+    def _resolve_fallback_credentials(
+        self, fb_provider: str, fb_config: dict
+    ) -> Optional[tuple]:
+        """Resolve credentials for a fallback provider.
+
+        Returns (api_key, base_url, api_mode) on success, or None on failure.
+        Handles three cases:
+          1. OAuth providers (openai-codex, nous) — call credential resolver
+          2. API-key providers (openrouter, zai, etc.) — read env var
+          3. Custom endpoints — use base_url + api_key_env from config
+        """
+        # ── 1. OAuth providers ────────────────────────────────────────
+        if fb_provider in self._FALLBACK_OAUTH_PROVIDERS:
+            resolver_name, api_mode = self._FALLBACK_OAUTH_PROVIDERS[fb_provider]
+            try:
+                import hermes_cli.auth as _auth
+                resolver = getattr(_auth, resolver_name)
+                creds = resolver()
+                return creds["api_key"], creds["base_url"], api_mode
+            except Exception as e:
+                logging.warning(
+                    "Fallback to %s failed (credential resolution): %s",
+                    fb_provider, e,
+                )
+                return None
+
+        # ── 2. API-key providers ──────────────────────────────────────
+        fb_key = (fb_config.get("api_key") or "").strip()
+        if not fb_key:
+            key_env = (fb_config.get("api_key_env") or "").strip()
+            if key_env:
+                fb_key = os.getenv(key_env, "")
+            elif fb_provider in self._FALLBACK_API_KEY_PROVIDERS:
+                for env_var in self._FALLBACK_API_KEY_PROVIDERS[fb_provider][1]:
+                    fb_key = os.getenv(env_var, "")
+                    if fb_key:
+                        break
+        if not fb_key:
+            logging.warning(
+                "Fallback model configured but no API key found for provider '%s'",
+                fb_provider,
+            )
+            return None
+
+        # ── 3. Resolve base URL ───────────────────────────────────────
+        fb_base_url = (fb_config.get("base_url") or "").strip()
+        if not fb_base_url and fb_provider in self._FALLBACK_API_KEY_PROVIDERS:
+            fb_base_url = self._FALLBACK_API_KEY_PROVIDERS[fb_provider][0]
+        if not fb_base_url:
+            fb_base_url = OPENROUTER_BASE_URL
+
+        return fb_key, fb_base_url, "chat_completions"
+
+    def _try_activate_fallback(self) -> bool:
+        """Switch to the configured fallback model/provider.
+
+        Called when the primary model is failing after retries.  Swaps the
+        OpenAI client, model slug, and provider in-place so the retry loop
+        can continue with the new backend.  One-shot: returns False if
+        already activated or not configured.
+        """
+        if self._fallback_activated or not self._fallback_model:
+            return False
+
+        fb = self._fallback_model
+        fb_provider = (fb.get("provider") or "").strip().lower()
+        fb_model = (fb.get("model") or "").strip()
+        if not fb_provider or not fb_model:
+            return False
+
+        resolved = self._resolve_fallback_credentials(fb_provider, fb)
+        if resolved is None:
+            return False
+        fb_key, fb_base_url, fb_api_mode = resolved
+
+        # Build new client
+        try:
+            client_kwargs = {"api_key": fb_key, "base_url": fb_base_url}
+            if "openrouter" in fb_base_url.lower():
+                client_kwargs["default_headers"] = {
+                    "HTTP-Referer": "https://github.com/NousResearch/hermes-agent",
+                    "X-OpenRouter-Title": "Hermes Agent",
+                    "X-OpenRouter-Categories": "productivity,cli-agent",
+                }
+            elif "api.kimi.com" in fb_base_url.lower():
+                client_kwargs["default_headers"] = {"User-Agent": "KimiCLI/1.0"}
+
+            self.client = OpenAI(**client_kwargs)
+            self._client_kwargs = client_kwargs
+            old_model = self.model
+            self.model = fb_model
+            self.provider = fb_provider
+            self.base_url = fb_base_url
+            self.api_mode = fb_api_mode
+            self._fallback_activated = True
+
+            # Re-evaluate prompt caching for the new provider/model
+            self._use_prompt_caching = (
+                "openrouter" in fb_base_url.lower()
+                and "claude" in fb_model.lower()
+            )
+
+            print(
+                f"{self.log_prefix}🔄 Primary model failed — switching to fallback: "
+                f"{fb_model} via {fb_provider}"
+            )
+            logging.info(
+                "Fallback activated: %s → %s (%s)",
+                old_model, fb_model, fb_provider,
+            )
+            return True
+        except Exception as e:
+            logging.error("Failed to activate fallback model: %s", e)
+            return False
+
+    # ── End provider fallback ──────────────────────────────────────────────
+
    def _build_api_kwargs(self, api_messages: list) -> dict:
        """Build the keyword arguments dict for the active API mode."""
        if self.api_mode == "codex_responses":
@@ -2503,6 +2665,8 @@ class AIAgent:
                    except (ValueError, Exception) as e:
                        logger.debug("Could not propagate title on compression: %s", e)
                self._session_db.update_system_prompt(self.session_id, new_system_prompt)
+                # Reset flush cursor — new session starts with no messages written
+                self._last_flushed_db_idx = 0
            except Exception as e:
                logger.debug("Session DB compression split failed: %s", e)

@@ -2526,7 +2690,6 @@ class AIAgent:
                        "tool_call_id": skipped_tc.id,
                    }
                    messages.append(skip_msg)
-                    self._log_msg_to_db(skip_msg)
                break

            function_name = tool_call.function.name
@@ -2542,6 +2705,8 @@ class AIAgent:
            except json.JSONDecodeError as e:
                logging.warning(f"Unexpected JSON error after validation: {e}")
                function_args = {}
+            if not isinstance(function_args, dict):
+                function_args = {}

            if not self.quiet_mode:
                args_str = json.dumps(function_args, ensure_ascii=False)
@@ -2555,6 +2720,18 @@ class AIAgent:
                except Exception as cb_err:
                    logging.debug(f"Tool progress callback error: {cb_err}")

+            # Checkpoint: snapshot working dir before file-mutating tools
+            if function_name in ("write_file", "patch") and self._checkpoint_mgr.enabled:
+                try:
+                    file_path = function_args.get("path", "")
+                    if file_path:
+                        work_dir = self._checkpoint_mgr.get_working_dir_for_path(file_path)
+                        self._checkpoint_mgr.ensure_checkpoint(
+                            work_dir, f"before {function_name}"
+                        )
+                except Exception:
+                    pass  # never block tool execution
+
            tool_start_time = time.time()

            if function_name == "todo":
@@ -2667,7 +2844,10 @@ class AIAgent:
                spinner.start()
                _spinner_result = None
                try:
-                    function_result = handle_function_call(function_name, function_args, effective_task_id)
+                    function_result = handle_function_call(
+                        function_name, function_args, effective_task_id,
+                        enabled_tools=list(self.valid_tool_names) if self.valid_tool_names else None,
+                    )
                    _spinner_result = function_result
                except Exception as tool_error:
                    function_result = f"Error executing tool '{function_name}': {tool_error}"
@@ -2678,7 +2858,10 @@ class AIAgent:
                    spinner.stop(cute_msg)
            else:
                try:
-                    function_result = handle_function_call(function_name, function_args, effective_task_id)
+                    function_result = handle_function_call(
+                        function_name, function_args, effective_task_id,
+                        enabled_tools=list(self.valid_tool_names) if self.valid_tool_names else None,
+                    )
                except Exception as tool_error:
                    function_result = f"Error executing tool '{function_name}': {tool_error}"
                    logger.error("handle_function_call raised for %s: %s", function_name, tool_error, exc_info=True)
@@ -2715,7 +2898,6 @@ class AIAgent:
                "tool_call_id": tool_call.id
            }
            messages.append(tool_msg)
-            self._log_msg_to_db(tool_msg)

            if not self.quiet_mode:
                response_preview = function_result[:self.log_prefix_chars] + "..." if len(function_result) > self.log_prefix_chars else function_result
@@ -2732,7 +2914,6 @@ class AIAgent:
                        "tool_call_id": skipped_tc.id
                    }
                    messages.append(skip_msg)
-                    self._log_msg_to_db(skip_msg)
                break

            if self.tool_delay > 0 and i < len(assistant_message.tool_calls):
@@ -2895,6 +3076,8 @@ class AIAgent:
        self._invalid_tool_retries = 0
        self._invalid_json_retries = 0
        self._empty_content_retries = 0
+        self._incomplete_scratchpad_retries = 0
+        self._codex_incomplete_retries = 0
        self._last_content_with_tools = None
        self._turns_since_memory = 0
        self._iters_since_skill = 0
@@ -2945,9 +3128,14 @@ class AIAgent:
            )
            self._iters_since_skill = 0

-        # Honcho prefetch: retrieve user context for system prompt injection
+        # Honcho prefetch: retrieve user context for system prompt injection.
+        # Only on the FIRST turn of a session (empty history).  On subsequent
+        # turns the model already has all prior context in its conversation
+        # history, and the Honcho context is baked into the stored system
+        # prompt — re-fetching it would change the system message and break
+        # Anthropic prompt caching.
        self._honcho_context = ""
-        if self._honcho and self._honcho_session_key:
+        if self._honcho and self._honcho_session_key and not conversation_history:
            try:
                self._honcho_context = self._honcho_prefetch(user_message)
            except Exception as e:
@@ -2956,7 +3144,6 @@ class AIAgent:
        # Add user message
        user_msg = {"role": "user", "content": user_message}
        messages.append(user_msg)
-        self._log_msg_to_db(user_msg)
        
        if not self.quiet_mode:
            print(f"💬 Starting conversation: '{user_message[:60]}{'...' if len(user_message) > 60 else ''}'")
@@ -2965,14 +3152,42 @@ class AIAgent:
        # Built once on first call, reused for all subsequent calls.
        # Only rebuilt after context compression events (which invalidate
        # the cache and reload memory from disk).
+        #
+        # For continuing sessions (gateway creates a fresh AIAgent per
+        # message), we load the stored system prompt from the session DB
+        # instead of rebuilding.  Rebuilding would pick up memory changes
+        # from disk that the model already knows about (it wrote them!),
+        # producing a different system prompt and breaking the Anthropic
+        # prefix cache.
        if self._cached_system_prompt is None:
-            self._cached_system_prompt = self._build_system_prompt(system_message)
-            # Store the system prompt snapshot in SQLite
-            if self._session_db:
+            stored_prompt = None
+            if conversation_history and self._session_db:
                try:
-                    self._session_db.update_system_prompt(self.session_id, self._cached_system_prompt)
-                except Exception as e:
-                    logger.debug("Session DB update_system_prompt failed: %s", e)
+                    session_row = self._session_db.get_session(self.session_id)
+                    if session_row:
+                        stored_prompt = session_row.get("system_prompt") or None
+                except Exception:
+                    pass  # Fall through to build fresh
+
+            if stored_prompt:
+                # Continuing session — reuse the exact system prompt from
+                # the previous turn so the Anthropic cache prefix matches.
+                self._cached_system_prompt = stored_prompt
+            else:
+                # First turn of a new session — build from scratch.
+                self._cached_system_prompt = self._build_system_prompt(system_message)
+                # Bake Honcho context into the prompt so it's stable for
+                # the entire session (not re-fetched per turn).
+                if self._honcho_context:
+                    self._cached_system_prompt = (
+                        self._cached_system_prompt + "\n\n" + self._honcho_context
+                    ).strip()
+                # Store the system prompt snapshot in SQLite
+                if self._session_db:
+                    try:
+                        self._session_db.update_system_prompt(self.session_id, self._cached_system_prompt)
+                    except Exception as e:
+                        logger.debug("Session DB update_system_prompt failed: %s", e)

        active_system_prompt = self._cached_system_prompt

@@ -3026,11 +3241,16 @@ class AIAgent:
        final_response = None
        interrupted = False
        codex_ack_continuations = 0
+        length_continue_retries = 0
+        truncated_response_prefix = ""
        
        # Clear any stale interrupt state at start
        self.clear_interrupt()
        
        while api_call_count < self.max_iterations and self.iteration_budget.remaining > 0:
+            # Reset per-turn checkpoint dedup so each iteration can take one snapshot
+            self._checkpoint_mgr.new_turn()
+
            # Check for interrupt request (e.g., user sent new message)
            if self._interrupt_requested:
                interrupted = True
@@ -3074,7 +3294,7 @@ class AIAgent:
            api_messages = []
            for msg in messages:
                api_msg = msg.copy()
-                
+
                # For ALL assistant messages, pass reasoning back to the API
                # This ensures multi-turn reasoning context is preserved
                if msg.get("role") == "assistant":
@@ -3082,7 +3302,7 @@ class AIAgent:
                    if reasoning_text:
                        # Add reasoning_content for API compatibility (Moonshot AI, Novita, OpenRouter)
                        api_msg["reasoning_content"] = reasoning_text
-                
+
                # Remove 'reasoning' field - it's for trajectory storage only
                # We've copied it to 'reasoning_content' for the API above
                if "reasoning" in api_msg:
@@ -3093,32 +3313,34 @@ class AIAgent:
                # Keep 'reasoning_details' - OpenRouter uses this for multi-turn reasoning context
                # The signature field helps maintain reasoning continuity
                api_messages.append(api_msg)
-            
+
            # Build the final system message: cached prompt + ephemeral system prompt.
            # The ephemeral part is appended here (not baked into the cached prompt)
            # so it stays out of the session DB and logs.
+            # Note: Honcho context is baked into _cached_system_prompt on the first
+            # turn and stored in the session DB, so it does NOT need to be injected
+            # here.  This keeps the system message identical across all turns in a
+            # session, maximizing Anthropic prompt cache hits.
            effective_system = active_system_prompt or ""
            if self.ephemeral_system_prompt:
                effective_system = (effective_system + "\n\n" + self.ephemeral_system_prompt).strip()
-            if self._honcho_context:
-                effective_system = (effective_system + "\n\n" + self._honcho_context).strip()
            if effective_system:
                api_messages = [{"role": "system", "content": effective_system}] + api_messages
-            
+
            # Inject ephemeral prefill messages right after the system prompt
            # but before conversation history. Same API-call-time-only pattern.
            if self.prefill_messages:
                sys_offset = 1 if effective_system else 0
                for idx, pfm in enumerate(self.prefill_messages):
                    api_messages.insert(sys_offset + idx, pfm.copy())
-            
+
            # Apply Anthropic prompt caching for Claude models via OpenRouter.
            # Auto-detected: if model name contains "claude" and base_url is OpenRouter,
            # inject cache_control breakpoints (system + last 3 messages) to reduce
            # input token costs by ~75% on multi-turn conversations.
            if self._use_prompt_caching:
                api_messages = apply_anthropic_cache_control(api_messages, cache_ttl=self._cache_ttl)
-            
+
            # Safety net: strip orphaned tool results / add stubs for missing
            # results before sending to the API.  The compressor handles this
            # during compression, but orphans can also sneak in from session
@@ -3141,9 +3363,13 @@ class AIAgent:
                # Animated thinking spinner in quiet mode
                face = random.choice(KawaiiSpinner.KAWAII_THINKING)
                verb = random.choice(KawaiiSpinner.THINKING_VERBS)
-                spinner_type = random.choice(['brain', 'sparkle', 'pulse', 'moon', 'star'])
-                thinking_spinner = KawaiiSpinner(f"{face} {verb}...", spinner_type=spinner_type)
-                thinking_spinner.start()
+                if self.thinking_callback:
+                    # CLI TUI mode: use prompt_toolkit widget instead of raw spinner
+                    self.thinking_callback(f"{face} {verb}...")
+                else:
+                    spinner_type = random.choice(['brain', 'sparkle', 'pulse', 'moon', 'star'])
+                    thinking_spinner = KawaiiSpinner(f"{face} {verb}...", spinner_type=spinner_type)
+                    thinking_spinner.start()
            
            # Log request details if verbose
            if self.verbose_logging:
@@ -3158,6 +3384,8 @@ class AIAgent:
            max_compression_attempts = 3
            codex_auth_retry_attempted = False
            nous_auth_retry_attempted = False
+            restart_with_compressed_messages = False
+            restart_with_length_continuation = False

            finish_reason = "stop"
            response = None  # Guard against UnboundLocalError if all retries fail
@@ -3180,6 +3408,8 @@ class AIAgent:
                    if thinking_spinner:
                        thinking_spinner.stop("")
                        thinking_spinner = None
+                    if self.thinking_callback:
+                        self.thinking_callback("")
                    
                    if not self.quiet_mode:
                        print(f"{self.log_prefix}⏱️  API call completed in {api_duration:.2f}s")
@@ -3220,6 +3450,8 @@ class AIAgent:
                        if thinking_spinner:
                            thinking_spinner.stop(f"(´;ω;`) oops, retrying...")
                            thinking_spinner = None
+                        if self.thinking_callback:
+                            self.thinking_callback("")
                        
                        # This is often rate limiting or provider returning malformed response
                        retry_count += 1
@@ -3252,6 +3484,10 @@ class AIAgent:
                        print(f"{self.log_prefix}   ⏱️  Response time: {api_duration:.2f}s (fast response often indicates rate limiting)")
                        
                        if retry_count >= max_retries:
+                            # Try fallback before giving up
+                            if self._try_activate_fallback():
+                                retry_count = 0
+                                continue
                            print(f"{self.log_prefix}❌ Max retries ({max_retries}) exceeded for invalid responses. Giving up.")
                            logging.error(f"{self.log_prefix}Invalid API response after {max_retries} retries.")
                            self._persist_session(messages, conversation_history)
@@ -3300,19 +3536,58 @@ class AIAgent:
                            finish_reason = "stop"
                    else:
                        finish_reason = response.choices[0].finish_reason
-                    
-                    # Handle "length" finish_reason - response was truncated
+
                    if finish_reason == "length":
                        print(f"{self.log_prefix}⚠️  Response truncated (finish_reason='length') - model hit max output tokens")
-                        
+
+                        if self.api_mode == "chat_completions":
+                            assistant_message = response.choices[0].message
+                            if not assistant_message.tool_calls:
+                                length_continue_retries += 1
+                                interim_msg = self._build_assistant_message(assistant_message, finish_reason)
+                                messages.append(interim_msg)
+                                if assistant_message.content:
+                                    truncated_response_prefix += assistant_message.content
+
+                                if length_continue_retries < 3:
+                                    print(
+                                        f"{self.log_prefix}↻ Requesting continuation "
+                                        f"({length_continue_retries}/3)..."
+                                    )
+                                    continue_msg = {
+                                        "role": "user",
+                                        "content": (
+                                            "[System: Your previous response was truncated by the output "
+                                            "length limit. Continue exactly where you left off. Do not "
+                                            "restart or repeat prior text. Finish the answer directly.]"
+                                        ),
+                                    }
+                                    messages.append(continue_msg)
+                                    self._session_messages = messages
+                                    self._save_session_log(messages)
+                                    restart_with_length_continuation = True
+                                    break
+
+                                partial_response = self._strip_think_blocks(truncated_response_prefix).strip()
+                                self._cleanup_task_resources(effective_task_id)
+                                self._persist_session(messages, conversation_history)
+                                return {
+                                    "final_response": partial_response or None,
+                                    "messages": messages,
+                                    "api_calls": api_call_count,
+                                    "completed": False,
+                                    "partial": True,
+                                    "error": "Response remained truncated after 3 continuation attempts",
+                                }
+
                        # If we have prior messages, roll back to last complete state
                        if len(messages) > 1:
                            print(f"{self.log_prefix}   ⏪ Rolling back to last complete assistant turn")
                            rolled_back_messages = self._get_messages_up_to_last_assistant(messages)
-                            
+
                            self._cleanup_task_resources(effective_task_id)
                            self._persist_session(messages, conversation_history)
-                            
+
                            return {
                                "final_response": None,
                                "messages": rolled_back_messages,
@@ -3385,6 +3660,8 @@ class AIAgent:
                    if thinking_spinner:
                        thinking_spinner.stop("")
                        thinking_spinner = None
+                    if self.thinking_callback:
+                        self.thinking_callback("")
                    api_elapsed = time.time() - api_start_time
                    print(f"{self.log_prefix}⚡ Interrupted during API call.")
                    self._persist_session(messages, conversation_history)
@@ -3397,6 +3674,8 @@ class AIAgent:
                    if thinking_spinner:
                        thinking_spinner.stop(f"(╥_╥) error, retrying...")
                        thinking_spinner = None
+                    if self.thinking_callback:
+                        self.thinking_callback("")

                    status_code = getattr(api_error, "status_code", None)
                    if (
@@ -3479,7 +3758,8 @@ class AIAgent:
                        if len(messages) < original_len:
                            print(f"{self.log_prefix}   🗜️  Compressed {original_len} → {len(messages)} messages, retrying...")
                            time.sleep(2)  # Brief pause between compression retries
-                            continue  # Retry with compressed messages
+                            restart_with_compressed_messages = True
+                            break
                        else:
                            print(f"{self.log_prefix}❌ Payload too large and cannot compress further.")
                            logging.error(f"{self.log_prefix}413 payload too large. Cannot compress further.")
@@ -3547,7 +3827,8 @@ class AIAgent:
                            if len(messages) < original_len:
                                print(f"{self.log_prefix}   🗜️  Compressed {original_len} → {len(messages)} messages, retrying...")
                            time.sleep(2)  # Brief pause between compression retries
-                            continue  # Retry with compressed messages or new tier
+                            restart_with_compressed_messages = True
+                            break
                        else:
                            # Can't compress further and already at minimum tier
                            print(f"{self.log_prefix}❌ Context length exceeded and cannot compress further.")
@@ -3576,6 +3857,11 @@ class AIAgent:
                    ])) and not is_context_length_error

                    if is_client_error:
+                        # Try fallback before aborting — a different provider
+                        # may not have the same issue (rate limit, auth, etc.)
+                        if self._try_activate_fallback():
+                            retry_count = 0
+                            continue
                        self._dump_api_request_debug(
                            api_kwargs, reason="non_retryable_client_error", error=api_error,
                        )
@@ -3593,6 +3879,10 @@ class AIAgent:
                        }

                    if retry_count >= max_retries:
+                        # Try fallback before giving up entirely
+                        if self._try_activate_fallback():
+                            retry_count = 0
+                            continue
                        print(f"{self.log_prefix}❌ Max retries ({max_retries}) exceeded. Giving up.")
                        logging.error(f"{self.log_prefix}API call failed after {max_retries} retries. Last error: {api_error}")
                        logging.error(f"{self.log_prefix}Request details - Messages: {len(api_messages)}, Approx tokens: {approx_tokens:,}")
@@ -3625,6 +3915,14 @@ class AIAgent:
            if interrupted:
                break

+            if restart_with_compressed_messages:
+                api_call_count -= 1
+                self.iteration_budget.refund()
+                continue
+
+            if restart_with_length_continuation:
+                continue
+
            # Guard: if all retries exhausted without a successful response
            # (e.g. repeated context-length errors that exhausted retry_count),
            # the `response` variable is still None. Break out cleanly.
@@ -3639,6 +3937,27 @@ class AIAgent:
                else:
                    assistant_message = response.choices[0].message
                
+                # Normalize content to string — some OpenAI-compatible servers
+                # (llama-server, etc.) return content as a dict or list instead
+                # of a plain string, which crashes downstream .strip() calls.
+                if assistant_message.content is not None and not isinstance(assistant_message.content, str):
+                    raw = assistant_message.content
+                    if isinstance(raw, dict):
+                        assistant_message.content = raw.get("text", "") or raw.get("content", "") or json.dumps(raw)
+                    elif isinstance(raw, list):
+                        # Multimodal content list — extract text parts
+                        parts = []
+                        for part in raw:
+                            if isinstance(part, str):
+                                parts.append(part)
+                            elif isinstance(part, dict) and part.get("type") == "text":
+                                parts.append(part.get("text", ""))
+                            elif isinstance(part, dict) and "text" in part:
+                                parts.append(str(part["text"]))
+                        assistant_message.content = "\n".join(parts)
+                    else:
+                        assistant_message.content = str(raw)
+
                # Handle assistant response
                if assistant_message.content and not self.quiet_mode:
                    print(f"{self.log_prefix}🤖 Assistant: {assistant_message.content[:100]}{'...' if len(assistant_message.content) > 100 else ''}")
@@ -3716,7 +4035,6 @@ class AIAgent:
                        )
                        if not duplicate_interim:
                            messages.append(interim_msg)
-                            self._log_msg_to_db(interim_msg)

                    if self._codex_incomplete_retries < 3:
                        if not self.quiet_mode:
@@ -3748,39 +4066,36 @@ class AIAgent:
                            logging.debug(f"Tool call: {tc.function.name} with args: {tc.function.arguments[:200]}...")
                    
                    # Validate tool call names - detect model hallucinations
+                    # Repair mismatched tool names before validating
+                    for tc in assistant_message.tool_calls:
+                        if tc.function.name not in self.valid_tool_names:
+                            repaired = self._repair_tool_call(tc.function.name)
+                            if repaired:
+                                print(f"{self.log_prefix}🔧 Auto-repaired tool name: '{tc.function.name}' -> '{repaired}'")
+                                tc.function.name = repaired
                    invalid_tool_calls = [
-                        tc.function.name for tc in assistant_message.tool_calls 
+                        tc.function.name for tc in assistant_message.tool_calls
                        if tc.function.name not in self.valid_tool_names
                    ]
-                    
                    if invalid_tool_calls:
-                        # Track retries for invalid tool calls
-                        if not hasattr(self, '_invalid_tool_retries'):
-                            self._invalid_tool_retries = 0
-                        self._invalid_tool_retries += 1
-                        
-                        invalid_preview = invalid_tool_calls[0][:80] + "..." if len(invalid_tool_calls[0]) > 80 else invalid_tool_calls[0]
-                        print(f"{self.log_prefix}⚠️  Invalid tool call detected: '{invalid_preview}'")
-                        print(f"{self.log_prefix}   Valid tools: {sorted(self.valid_tool_names)}")
-                        
-                        if self._invalid_tool_retries < 3:
-                            print(f"{self.log_prefix}🔄 Retrying API call ({self._invalid_tool_retries}/3)...")
-                            # Don't add anything to messages, just retry the API call
-                            continue
-                        else:
-                            print(f"{self.log_prefix}❌ Max retries (3) for invalid tool calls exceeded. Stopping as partial.")
-                            # Return partial result - don't include the bad tool call in messages
-                            self._invalid_tool_retries = 0
-                            self._persist_session(messages, conversation_history)
-                            return {
-                                "final_response": None,
-                                "messages": messages,
-                                "api_calls": api_call_count,
-                                "completed": False,
-                                "partial": True,
-                                "error": f"Model generated invalid tool call: {invalid_preview}"
-                            }
-                    
+                        # Return helpful error to model — model can self-correct next turn
+                        available = ", ".join(sorted(self.valid_tool_names))
+                        invalid_name = invalid_tool_calls[0]
+                        invalid_preview = invalid_name[:80] + "..." if len(invalid_name) > 80 else invalid_name
+                        print(f"{self.log_prefix}⚠️  Unknown tool '{invalid_preview}' — sending error to model for self-correction")
+                        assistant_msg = self._build_assistant_message(assistant_message, finish_reason)
+                        messages.append(assistant_msg)
+                        for tc in assistant_message.tool_calls:
+                            if tc.function.name not in self.valid_tool_names:
+                                content = f"Tool '{tc.function.name}' does not exist. Available tools: {available}"
+                            else:
+                                content = f"Skipped: another tool call in this turn used an invalid name. Please retry this tool call."
+                            messages.append({
+                                "role": "tool",
+                                "tool_call_id": tc.id,
+                                "content": content,
+                            })
+                        continue
                    # Reset retry counter on successful tool call validation
                    if hasattr(self, '_invalid_tool_retries'):
                        self._invalid_tool_retries = 0
@@ -3824,7 +4139,6 @@ class AIAgent:
                            )
                            recovery_dict = {"role": "user", "content": recovery_msg}
                            messages.append(recovery_dict)
-                            self._log_msg_to_db(recovery_dict)
                            continue
                    
                    # Reset retry counter on successful JSON validation
@@ -3846,7 +4160,6 @@ class AIAgent:
                                print(f"  ┊ 💬 {clean}")
                    
                    messages.append(assistant_msg)
-                    self._log_msg_to_db(assistant_msg)
                    
                    self._execute_tool_calls(assistant_message, messages, effective_task_id)

@@ -3947,7 +4260,6 @@ class AIAgent:
                                "finish_reason": finish_reason,
                            }
                            messages.append(empty_msg)
-                            self._log_msg_to_db(empty_msg)
                            
                            self._cleanup_task_resources(effective_task_id)
                            self._persist_session(messages, conversation_history)
@@ -3978,7 +4290,6 @@ class AIAgent:
                        codex_ack_continuations += 1
                        interim_msg = self._build_assistant_message(assistant_message, "incomplete")
                        messages.append(interim_msg)
-                        self._log_msg_to_db(interim_msg)

                        continue_msg = {
                            "role": "user",
@@ -3988,12 +4299,14 @@ class AIAgent:
                            ),
                        }
                        messages.append(continue_msg)
-                        self._log_msg_to_db(continue_msg)
                        self._session_messages = messages
                        self._save_session_log(messages)
                        continue

                    codex_ack_continuations = 0
+
+                    if truncated_response_prefix:
+                        final_response = truncated_response_prefix + final_response
                    
                    # Strip <think> blocks from user-facing response (keep raw in messages for trajectory)
                    final_response = self._strip_think_blocks(final_response).strip()
@@ -4001,7 +4314,6 @@ class AIAgent:
                    final_msg = self._build_assistant_message(assistant_message, finish_reason)
                    
                    messages.append(final_msg)
-                    self._log_msg_to_db(final_msg)
                    
                    if not self.quiet_mode:
                        print(f"🎉 Conversation completed after {api_call_count} OpenAI-compatible API call(s)")
@@ -4038,7 +4350,6 @@ class AIAgent:
                                    "content": f"Error executing tool: {error_msg}",
                                }
                                messages.append(err_msg)
-                                self._log_msg_to_db(err_msg)
                        pending_handled = True
                    break
                
@@ -4051,7 +4362,6 @@ class AIAgent:
                        "content": f"[System error during processing: {error_msg}]",
                    }
                    messages.append(sys_err_msg)
-                    self._log_msg_to_db(sys_err_msg)
                
                # If we're near the limit, break to avoid infinite loops
                if api_call_count >= self.max_iterations - 1:
--- a/scripts/install.sh
+++ b/scripts/install.sh
@@ -492,9 +492,23 @@ install_system_packages() {
                        return 0
                    fi
                fi
+            elif [ -e /dev/tty ]; then
+                # Non-interactive (e.g. curl | bash) but a terminal is available.
+                # Read the prompt from /dev/tty (same approach the setup wizard uses).
+                echo ""
+                log_info "Installing ${description} requires sudo."
+                read -p "Install? [Y/n] " -n 1 -r < /dev/tty
+                echo
+                if [[ $REPLY =~ ^[Yy]$ ]] || [[ -z $REPLY ]]; then
+                    if sudo DEBIAN_FRONTEND=noninteractive NEEDRESTART_MODE=a $install_cmd < /dev/tty; then
+                        [ "$need_ripgrep" = true ] && HAS_RIPGREP=true && log_success "ripgrep installed"
+                        [ "$need_ffmpeg" = true ]  && HAS_FFMPEG=true  && log_success "ffmpeg installed"
+                        return 0
+                    fi
+                fi
            else
-                log_warn "Non-interactive mode: cannot prompt for sudo password"
-                log_info "Install missing packages manually: sudo $install_cmd"
+                log_warn "Non-interactive mode and no terminal available — cannot install system packages"
+                log_info "Install manually after setup completes: sudo $install_cmd"
            fi
        fi
    fi
--- a/skills/creative/DESCRIPTION.md
+++ b/skills/creative/DESCRIPTION.md
@@ -0,0 +1,3 @@
+---
+description: Creative content generation — ASCII art, hand-drawn style diagrams, and visual design tools.
+---
--- a/skills/creative/ascii-video/SKILL.md
+++ b/skills/creative/ascii-video/SKILL.md
@@ -0,0 +1,250 @@
+---
+name: ascii-video
+description: "Production pipeline for ASCII art video — any format. Converts video/audio/images/generative input into colored ASCII character video output (MP4, GIF, image sequence). Covers: video-to-ASCII conversion, audio-reactive music visualizers, generative ASCII art animations, hybrid video+audio reactive, text/lyrics overlays, real-time terminal rendering. Use when users request: ASCII video, text art video, terminal-style video, character art animation, retro text visualization, audio visualizer in ASCII, converting video to ASCII art, matrix-style effects, or any animated ASCII output."
+---
+
+# ASCII Video Production Pipeline
+
+Full production pipeline for rendering any content as colored ASCII character video.
+
+## Modes
+
+| Mode | Input | Output | Read |
+|------|-------|--------|------|
+| **Video-to-ASCII** | Video file | ASCII recreation of source footage | `references/inputs.md` § Video Sampling |
+| **Audio-reactive** | Audio file | Generative visuals driven by audio features | `references/inputs.md` § Audio Analysis |
+| **Generative** | None (or seed params) | Procedural ASCII animation | `references/effects.md` |
+| **Hybrid** | Video + audio | ASCII video with audio-reactive overlays | Both input refs |
+| **Lyrics/text** | Audio + text/SRT | Timed text with visual effects | `references/inputs.md` § Text/Lyrics |
+| **TTS narration** | Text quotes + TTS API | Narrated testimonial/quote video with typed text | `references/inputs.md` § TTS Integration |
+
+## Stack
+
+Single self-contained Python script per project. No GPU.
+
+| Layer | Tool | Purpose |
+|-------|------|---------|
+| Core | Python 3.10+, NumPy | Math, array ops, vectorized effects |
+| Signal | SciPy | FFT, peak detection (audio modes only) |
+| Imaging | Pillow (PIL) | Font rasterization, video frame decoding, image I/O |
+| Video I/O | ffmpeg (CLI) | Decode input, encode output segments, mux audio, mix tracks |
+| Parallel | concurrent.futures / multiprocessing | N workers for batch/clip rendering |
+| TTS | ElevenLabs API (or similar) | Generate narration clips for quote/testimonial videos |
+| Optional | OpenCV | Video frame sampling, edge detection, optical flow |
+
+## Pipeline Architecture (v2)
+
+Every mode follows the same 6-stage pipeline. See `references/architecture.md` for implementation details, `references/scenes.md` for scene protocol, and `references/composition.md` for multi-grid composition and tonemap.
+
+```
+┌─────────┐   ┌──────────┐   ┌───────────┐   ┌──────────┐   ┌─────────┐   ┌────────┐
+│ 1.INPUT  │→│ 2.ANALYZE │→│ 3.SCENE_FN │→│ 4.TONEMAP │→│ 5.SHADE  │→│ 6.ENCODE│
+│ load src │  │ features  │  │ → canvas   │  │ normalize │  │ post-fx  │  │ → video │
+└─────────┘   └──────────┘   └───────────┘   └──────────┘   └─────────┘   └────────┘
+```
+
+1. **INPUT** — Load/decode source material (video frames, audio samples, images, or nothing)
+2. **ANALYZE** — Extract per-frame features (audio bands, video luminance/edges, motion vectors)
+3. **SCENE_FN** — Scene function renders directly to pixel canvas (`uint8 H,W,3`). May internally compose multiple character grids via `_render_vf()` + pixel blend modes. See `references/composition.md`
+4. **TONEMAP** — Percentile-based adaptive brightness normalization with per-scene gamma. Replaces linear brightness multipliers. See `references/composition.md` § Adaptive Tonemap
+5. **SHADE** — Apply post-processing `ShaderChain` + `FeedbackBuffer`. See `references/shaders.md`
+6. **ENCODE** — Pipe raw RGB frames to ffmpeg for H.264/GIF encoding
+
+## Creative Direction
+
+**Every project should look and feel different.** The references provide a vocabulary of building blocks — don't copy them verbatim. Combine, modify, and invent.
+
+### Aesthetic Dimensions to Vary
+
+| Dimension | Options | Reference |
+|-----------|---------|-----------|
+| **Character palette** | Density ramps, block elements, symbols, scripts (katakana, Greek, runes, braille), dots, project-specific | `architecture.md` § Character Palettes |
+| **Color strategy** | HSV (angle/distance/time/value mapped), discrete RGB palettes, monochrome, complementary, triadic, temperature | `architecture.md` § Color System |
+| **Color tint** | Warm, cool, amber, matrix green, neon pink, sepia, ice, blood, void, sunset | `shaders.md` § Color Grade |
+| **Background texture** | Sine fields, noise, smooth noise, cellular/voronoi, video source | `effects.md` § Background Fills |
+| **Primary effects** | Rings, spirals, tunnel, vortex, waves, interference, aurora, ripple, fire | `effects.md` § Radial / Wave / Fire |
+| **Particles** | Energy sparks, snow, rain, bubbles, runes, binary data, orbits, gravity wells | `effects.md` § Particle Systems |
+| **Shader mood** | Retro CRT, clean modern, glitch art, cinematic, dreamy, harsh industrial, psychedelic | `shaders.md` § Design Philosophy |
+| **Grid density** | xs(8px) through xxl(40px), mixed per layer | `architecture.md` § Grid System |
+| **Font** | Menlo, Monaco, Courier, SF Mono, JetBrains Mono, Fira Code, IBM Plex | `architecture.md` § Font Selection |
+| **Mirror mode** | None, horizontal, vertical, quad, diagonal, kaleidoscope | `shaders.md` § Mirror Effects |
+| **Transition style** | Crossfade, wipe (directional/radial), dissolve, glitch cut | `shaders.md` § Transitions |
+
+### Per-Section Variation
+
+Never use the same config for the entire video. For each section/scene/quote:
+- Choose a **different background effect** (or compose 2-3)
+- Choose a **different character palette** (match the mood)
+- Choose a **different color strategy** (or at minimum a different hue)
+- Vary **shader intensity** (more bloom during peaks, more grain during quiet)
+- Use **different particle types** if particles are active
+
+### Project-Specific Invention
+
+For every project, invent at least one of:
+- A custom character palette matching the theme
+- A custom background effect (combine/modify existing ones)
+- A custom color palette (discrete RGB set matching the brand/mood)
+- A custom particle character set
+
+## Workflow
+
+### Step 1: Determine Mode and Gather Requirements
+
+Establish with user:
+- **Input source** — file path, format, duration
+- **Mode** — which of the 6 modes above
+- **Sections** — time-mapped style changes (timestamps → effect names)
+- **Resolution** — default 1920x1080 @ 24fps; GIFs typically 640x360 @ 15fps
+- **Style direction** — dense/sparse, bright/dark, chaotic/minimal, color palette
+- **Text/branding** — easter eggs, overlays, credits, themed character sets
+- **Output format** — MP4 (default), GIF, PNG sequence
+
+### Step 2: Detect Hardware and Set Quality
+
+Before building the script, detect the user's hardware and set appropriate defaults. See `references/optimization.md` § Hardware Detection.
+
+```python
+hw = detect_hardware()
+profile = quality_profile(hw, target_duration, user_quality_pref)
+log(f"Hardware: {hw['cpu_count']} cores, {hw['mem_gb']:.1f}GB RAM")
+log(f"Render: {profile['vw']}x{profile['vh']} @{profile['fps']}fps, {profile['workers']} workers")
+```
+
+Never hardcode worker counts, resolution, or CRF. Always detect and adapt.
+
+### Step 3: Build the Script
+
+Write as a single Python file. Major components:
+
+1. **Hardware detection + quality profile** — see `references/optimization.md`
+2. **Input loader** — mode-dependent; see `references/inputs.md`
+3. **Feature analyzer** — audio FFT, video luminance, or pass-through
+4. **Grid + renderer** — multi-density character grids with bitmap cache; `_render_vf()` helper for value/hue field → canvas
+5. **Character palettes** — multiple palettes chosen per project theme; see `references/architecture.md`
+6. **Color system** — HSV + discrete RGB palettes as needed; see `references/architecture.md`
+7. **Scene functions** — each returns `canvas (uint8 H,W,3)` directly. May compose multiple grids internally via pixel blend modes. See `references/scenes.md` + `references/composition.md`
+8. **Tonemap** — adaptive brightness normalization with per-scene gamma; see `references/composition.md`
+9. **Shader pipeline** — `ShaderChain` + `FeedbackBuffer` per-section config; see `references/shaders.md`
+10. **Scene table + dispatcher** — maps time ranges to scene functions + shader/feedback configs; see `references/scenes.md`
+11. **Parallel encoder** — N-worker batch clip rendering with ffmpeg pipes
+12. **Main** — orchestrate full pipeline
+
+### Step 4: Handle Critical Bugs
+
+#### Font Cell Height (macOS Pillow)
+
+`textbbox()` returns wrong height. Use `font.getmetrics()`:
+
+```python
+ascent, descent = font.getmetrics()
+cell_height = ascent + descent  # correct
+```
+
+#### ffmpeg Pipe Deadlock
+
+Never use `stderr=subprocess.PIPE` with long-running ffmpeg. Redirect to file:
+
+```python
+stderr_fh = open(err_path, "w")
+pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.DEVNULL, stderr=stderr_fh)
+```
+
+#### Brightness — Use `tonemap()`, Not Linear Multipliers
+
+ASCII on black is inherently dark. This is the #1 visual issue. **Do NOT use linear `* N` brightness multipliers** — they clip highlights and wash out the image. Instead, use the **adaptive tonemap** function from `references/composition.md`:
+
+```python
+def tonemap(canvas, gamma=0.75):
+    """Percentile-based adaptive normalization + gamma. Replaces all brightness multipliers."""
+    f = canvas.astype(np.float32)
+    lo = np.percentile(f, 1)          # black point (1st percentile)
+    hi = np.percentile(f, 99.5)       # white point (99.5th percentile)
+    if hi - lo < 1: hi = lo + 1
+    f = (f - lo) / (hi - lo)
+    f = np.clip(f, 0, 1) ** gamma     # gamma < 1 = brighter mids
+    return (f * 255).astype(np.uint8)
+```
+
+Pipeline ordering: `scene_fn() → tonemap() → FeedbackBuffer → ShaderChain → ffmpeg`
+
+Per-scene gamma overrides for destructive effects:
+- Default: `gamma=0.75`
+- Solarize scenes: `gamma=0.55` (solarize darkens above-threshold pixels)
+- Posterize scenes: `gamma=0.50` (quantization loses brightness range)
+- Already-bright scenes: `gamma=0.85`
+
+Additional brightness best practices:
+- Dense animated backgrounds — never flat black, always fill the grid
+- Vignette minimum clamped to 0.15 (not 0.12)
+- Bloom threshold lowered to 130 (not 170) so more pixels contribute to glow
+- Use `screen` blend mode (not `overlay`) when compositing dark ASCII layers — overlay squares dark values: `2 * 0.12 * 0.12 = 0.03`
+
+#### Font Compatibility
+
+Not all Unicode characters render in all fonts. Validate palettes at init:
+```python
+for c in palette:
+    img = Image.new("L", (20, 20), 0)
+    ImageDraw.Draw(img).text((0, 0), c, fill=255, font=font)
+    if np.array(img).max() == 0:
+        log(f"WARNING: char '{c}' (U+{ord(c):04X}) not in font, removing from palette")
+```
+
+### Step 4b: Per-Clip Architecture (for segmented videos)
+
+When the video has discrete segments (quotes, scenes, chapters), render each as a separate clip file. This enables:
+- Re-rendering individual clips without touching the rest (`--clip q05`)
+- Faster iteration on specific sections
+- Easy reordering or trimming in post
+
+```python
+segments = [
+    {"id": "intro", "start": 0.0, "end": 5.0, "type": "intro"},
+    {"id": "q00", "start": 5.0, "end": 12.0, "type": "quote", "qi": 0, ...},
+    {"id": "t00", "start": 12.0, "end": 13.5, "type": "transition", ...},
+    {"id": "outro", "start": 208.0, "end": 211.6, "type": "outro"},
+]
+
+from concurrent.futures import ProcessPoolExecutor, as_completed
+with ProcessPoolExecutor(max_workers=hw["workers"]) as pool:
+    futures = {pool.submit(render_clip, seg, features, path): seg["id"]
+               for seg, path in clip_args}
+    for fut in as_completed(futures):
+        fut.result()
+```
+
+CLI: `--clip q00 t00 q01` to re-render specific clips, `--list` to show segments, `--skip-render` to re-stitch only.
+
+### Step 5: Render and Iterate
+
+Performance targets per frame:
+
+| Component | Budget |
+|-----------|--------|
+| Feature extraction | 1-5ms |
+| Effect function | 2-15ms |
+| Character render | 80-150ms (bottleneck) |
+| Shader pipeline | 5-25ms |
+| **Total** | ~100-200ms/frame |
+
+**Fast iteration**: render single test frames to check brightness/layout before full render:
+```python
+canvas = render_single_frame(frame_index, features, renderer)
+Image.fromarray(canvas).save("test.png")
+```
+
+**Brightness verification**: sample 5-10 frames across video, check `mean > 8` for ASCII content.
+
+## References
+
+| File | Contents |
+|------|----------|
+| `references/architecture.md` | Grid system, font selection, character palettes (library of 20+), color system (HSV + discrete RGB), `_render_vf()` helper, compositing, v2 effect function contract |
+| `references/inputs.md` | All input sources: audio analysis, video sampling, image conversion, text/lyrics, TTS integration (ElevenLabs, voice assignment, audio mixing) |
+| `references/effects.md` | Effect building blocks: 12 value field generators (`vf_sinefield` through `vf_noise_static`), 8 hue field generators (`hf_fixed` through `hf_plasma`), radial/wave/fire effects, particles, composing guide |
+| `references/shaders.md` | 38 shader implementations (geometry, channel, color, glow, noise, pattern, tone, glitch, mirror), `ShaderChain` class, full `_apply_shader_step()` dispatch, audio-reactive scaling, transitions, tint presets |
+| `references/composition.md` | **v2 core**: pixel blend modes (20 modes with implementations), multi-grid composition, `_render_vf()` helper, adaptive `tonemap()`, per-scene gamma, `FeedbackBuffer` with spatial transforms, `PixelBlendStack` |
+| `references/scenes.md` | **v2 scene protocol**: scene function contract, `Renderer` class, `SCENES` table structure, `render_clip()` loop, beat-synced cutting, parallel rendering + pickling constraints, 4 complete scene examples, scene design checklist |
+| `references/troubleshooting.md` | NumPy broadcasting traps, blend mode pitfalls, multiprocessing/pickling issues, brightness diagnostics, ffmpeg deadlocks, font issues, performance bottlenecks, common mistakes |
+| `references/optimization.md` | Hardware detection, adaptive quality profiles (draft/preview/production/max), CLI integration, vectorized effect patterns, parallel rendering, memory management |
--- a/skills/creative/ascii-video/references/architecture.md
+++ b/skills/creative/ascii-video/references/architecture.md
@@ -0,0 +1,528 @@
+# Architecture Reference
+
+## Grid System
+
+### Multi-Density Grids
+
+Pre-initialize multiple grid sizes. Switch per section for visual variety.
+
+| Key | Font Size | Grid (1920x1080) | Use |
+|-----|-----------|-------------------|-----|
+| xs | 8 | 400x108 | Ultra-dense data fields |
+| sm | 10 | 320x83 | Dense detail, rain, starfields |
+| md | 16 | 192x56 | Default balanced, transitions |
+| lg | 20 | 160x45 | Quote/lyric text (readable at 1080p) |
+| xl | 24 | 137x37 | Short quotes, large titles |
+| xxl | 40 | 80x22 | Giant text, minimal |
+
+**Grid sizing for text-heavy content**: When displaying readable text (quotes, lyrics, testimonials), use 20px (`lg`) as the primary grid. This gives 160 columns -- plenty for lines up to ~50 chars centered. For very short quotes (< 60 chars, <= 3 lines), 24px (`xl`) makes them more impactful. Only init the grids you actually use -- each grid pre-rasterizes all characters which costs ~0.3-0.5s.
+
+Grid dimensions: `cols = VW // cell_width`, `rows = VH // cell_height`.
+
+### Font Selection
+
+Don't hardcode a single font. Choose fonts to match the project's mood. Monospace fonts are required for grid alignment but vary widely in personality:
+
+| Font | Personality | Platform |
+|------|-------------|----------|
+| Menlo | Clean, neutral, Apple-native | macOS |
+| Monaco | Retro terminal, compact | macOS |
+| Courier New | Classic typewriter, wide | Cross-platform |
+| SF Mono | Modern, tight spacing | macOS |
+| Consolas | Windows native, clean | Windows |
+| JetBrains Mono | Developer, ligature-ready | Install |
+| Fira Code | Geometric, modern | Install |
+| IBM Plex Mono | Corporate, authoritative | Install |
+| Source Code Pro | Adobe, balanced | Install |
+
+**Font detection at init**: probe available fonts and fall back gracefully:
+
+```python
+import platform
+
+def find_font(preferences):
+    """Try fonts in order, return first that exists."""
+    for name, path in preferences:
+        if os.path.exists(path):
+            return path
+    raise FileNotFoundError(f"No monospace font found. Tried: {[p for _,p in preferences]}")
+
+FONT_PREFS_MACOS = [
+    ("Menlo", "/System/Library/Fonts/Menlo.ttc"),
+    ("Monaco", "/System/Library/Fonts/Monaco.ttf"),
+    ("SF Mono", "/System/Library/Fonts/SFNSMono.ttf"),
+    ("Courier", "/System/Library/Fonts/Courier.ttc"),
+]
+FONT_PREFS_LINUX = [
+    ("DejaVu Sans Mono", "/usr/share/fonts/truetype/dejavu/DejaVuSansMono.ttf"),
+    ("Liberation Mono", "/usr/share/fonts/truetype/liberation/LiberationMono-Regular.ttf"),
+    ("Noto Sans Mono", "/usr/share/fonts/truetype/noto/NotoSansMono-Regular.ttf"),
+    ("Ubuntu Mono", "/usr/share/fonts/truetype/ubuntu/UbuntuMono-R.ttf"),
+]
+FONT_PREFS = FONT_PREFS_MACOS if platform.system() == "Darwin" else FONT_PREFS_LINUX
+```
+
+**Multi-font rendering**: use different fonts for different layers (e.g., monospace for background, a bolder variant for overlay text). Each GridLayer owns its own font:
+
+```python
+grid_bg = GridLayer(find_font(FONT_PREFS), 16)       # background
+grid_text = GridLayer(find_font(BOLD_PREFS), 20)      # readable text
+```
+
+### Collecting All Characters
+
+Before initializing grids, gather all characters that need bitmap pre-rasterization:
+
+```python
+all_chars = set()
+for pal in [PAL_DEFAULT, PAL_DENSE, PAL_BLOCKS, PAL_RUNE, PAL_KATA,
+            PAL_GREEK, PAL_MATH, PAL_DOTS, PAL_BRAILLE, PAL_STARS,
+            PAL_BINARY, PAL_MUSIC, PAL_BOX, PAL_CIRCUIT, PAL_ARROWS,
+            PAL_HERMES]:  # ... all palettes used in project
+    all_chars.update(pal)
+# Add any overlay text characters
+all_chars.update("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789 .,-:;!?/|")
+all_chars.discard(" ")  # space is never rendered
+```
+
+### GridLayer Initialization
+
+Each grid pre-computes coordinate arrays for vectorized effect math:
+
+```python
+class GridLayer:
+    def __init__(self, font_path, font_size):
+        self.font = ImageFont.truetype(font_path, font_size)
+        asc, desc = self.font.getmetrics()
+        bbox = self.font.getbbox("M")
+        self.cw = bbox[2] - bbox[0]  # character cell width
+        self.ch = asc + desc  # CRITICAL: not textbbox height
+
+        self.cols = VW // self.cw
+        self.rows = VH // self.ch
+        self.ox = (VW - self.cols * self.cw) // 2  # centering
+        self.oy = (VH - self.rows * self.ch) // 2
+
+        # Index arrays
+        self.rr = np.arange(self.rows, dtype=np.float32)[:, None]
+        self.cc = np.arange(self.cols, dtype=np.float32)[None, :]
+
+        # Polar coordinates (aspect-corrected)
+        cx, cy = self.cols / 2.0, self.rows / 2.0
+        asp = self.cw / self.ch
+        self.dx = self.cc - cx
+        self.dy = (self.rr - cy) * asp
+        self.dist = np.sqrt(self.dx**2 + self.dy**2)
+        self.angle = np.arctan2(self.dy, self.dx)
+
+        # Normalized (0-1 range) -- for distance falloff
+        self.dx_n = (self.cc - cx) / max(self.cols, 1)
+        self.dy_n = (self.rr - cy) / max(self.rows, 1) * asp
+        self.dist_n = np.sqrt(self.dx_n**2 + self.dy_n**2)
+
+        # Pre-rasterize all characters to float32 bitmaps
+        self.bm = {}
+        for c in all_chars:
+            img = Image.new("L", (self.cw, self.ch), 0)
+            ImageDraw.Draw(img).text((0, 0), c, fill=255, font=self.font)
+            self.bm[c] = np.array(img, dtype=np.float32) / 255.0
+```
+
+### Character Render Loop
+
+The bottleneck. Composites pre-rasterized bitmaps onto pixel canvas:
+
+```python
+def render(self, chars, colors, canvas=None):
+    if canvas is None:
+        canvas = np.zeros((VH, VW, 3), dtype=np.uint8)
+    for row in range(self.rows):
+        y = self.oy + row * self.ch
+        if y + self.ch > VH: break
+        for col in range(self.cols):
+            c = chars[row, col]
+            if c == " ": continue
+            x = self.ox + col * self.cw
+            if x + self.cw > VW: break
+            a = self.bm[c]  # float32 bitmap
+            canvas[y:y+self.ch, x:x+self.cw] = np.maximum(
+                canvas[y:y+self.ch, x:x+self.cw],
+                (a[:, :, None] * colors[row, col]).astype(np.uint8))
+    return canvas
+```
+
+Use `np.maximum` for additive blending (brighter chars overwrite dimmer ones, never darken).
+
+### Multi-Layer Rendering
+
+Render multiple grids onto the same canvas for depth:
+
+```python
+canvas = np.zeros((VH, VW, 3), dtype=np.uint8)
+canvas = grid_lg.render(bg_chars, bg_colors, canvas)   # background layer
+canvas = grid_md.render(main_chars, main_colors, canvas)  # main layer
+canvas = grid_sm.render(detail_chars, detail_colors, canvas)  # detail overlay
+```
+
+---
+
+## Character Palettes
+
+### Design Principles
+
+Character palettes are the primary visual texture of ASCII video. They control not just brightness mapping but the entire visual feel. Design palettes intentionally:
+
+- **Visual weight**: characters sorted by the amount of ink/pixels they fill. Space is always index 0.
+- **Coherence**: characters within a palette should belong to the same visual family.
+- **Density curve**: the brightness-to-character mapping is nonlinear. Dense palettes (many chars) give smoother gradients; sparse palettes (5-8 chars) give posterized/graphic looks.
+- **Rendering compatibility**: every character in the palette must exist in the font. Test at init and remove missing glyphs.
+
+### Palette Library
+
+Organized by visual family. Mix and match per project -- don't default to PAL_DEFAULT for everything.
+
+#### Density / Brightness Palettes
+```python
+PAL_DEFAULT  = " .`'-:;!><=+*^~?/|(){}[]#&$@%"       # classic ASCII art
+PAL_DENSE    = " .:;+=xX$#@\u2588"                          # simple 11-level ramp
+PAL_MINIMAL  = " .:-=+#@"                               # 8-level, graphic
+PAL_BINARY   = " \u2588"                                      # 2-level, extreme contrast
+PAL_GRADIENT = " \u2591\u2592\u2593\u2588"                              # 4-level block gradient
+```
+
+#### Unicode Block Elements
+```python
+PAL_BLOCKS   = " \u2591\u2592\u2593\u2588\u2584\u2580\u2590\u258c"                 # standard blocks
+PAL_BLOCKS_EXT = " \u2596\u2597\u2598\u2599\u259a\u259b\u259c\u259d\u259e\u259f\u2591\u2592\u2593\u2588"  # quadrant blocks (more detail)
+PAL_SHADE    = " \u2591\u2592\u2593\u2588\u2587\u2586\u2585\u2584\u2583\u2582\u2581"          # vertical fill progression
+```
+
+#### Symbolic / Thematic
+```python
+PAL_MATH     = " \u00b7\u2218\u2219\u2022\u00b0\u00b1\u2213\u00d7\u00f7\u2248\u2260\u2261\u2264\u2265\u221e\u222b\u2211\u220f\u221a\u2207\u2202\u2206\u03a9"    # math symbols
+PAL_BOX      = " \u2500\u2502\u250c\u2510\u2514\u2518\u251c\u2524\u252c\u2534\u253c\u2550\u2551\u2554\u2557\u255a\u255d\u2560\u2563\u2566\u2569\u256c"          # box drawing
+PAL_CIRCUIT  = " .\u00b7\u2500\u2502\u250c\u2510\u2514\u2518\u253c\u25cb\u25cf\u25a1\u25a0\u2206\u2207\u2261"                 # circuit board
+PAL_RUNE     = " .\u16a0\u16a2\u16a6\u16b1\u16b7\u16c1\u16c7\u16d2\u16d6\u16da\u16de\u16df"                   # elder futhark runes
+PAL_ALCHEMIC = " \u2609\u263d\u2640\u2642\u2643\u2644\u2645\u2646\u2647\u2648\u2649\u264a\u264b"            # planetary/alchemical symbols
+PAL_ZODIAC   = " \u2648\u2649\u264a\u264b\u264c\u264d\u264e\u264f\u2650\u2651\u2652\u2653"            # zodiac
+PAL_ARROWS   = " \u2190\u2191\u2192\u2193\u2194\u2195\u2196\u2197\u2198\u2199\u21a9\u21aa\u21bb\u27a1"             # directional arrows
+PAL_MUSIC    = " \u266a\u266b\u266c\u2669\u266d\u266e\u266f\u25cb\u25cf"                       # musical notation
+```
+
+#### Script / Writing System
+```python
+PAL_KATA     = " \u00b7\uff66\uff67\uff68\uff69\uff6a\uff6b\uff6c\uff6d\uff6e\uff6f\uff70\uff71\uff72\uff73\uff74\uff75\uff76\uff77"          # katakana halfwidth (matrix rain)
+PAL_GREEK    = " \u03b1\u03b2\u03b3\u03b4\u03b5\u03b6\u03b7\u03b8\u03b9\u03ba\u03bb\u03bc\u03bd\u03be\u03c0\u03c1\u03c3\u03c4\u03c6\u03c8\u03c9"    # Greek lowercase
+PAL_CYRILLIC = " \u0430\u0431\u0432\u0433\u0434\u0435\u0436\u0437\u0438\u043a\u043b\u043c\u043d\u043e\u043f\u0440\u0441\u0442\u0443\u0444\u0445\u0446\u0447\u0448"  # Cyrillic lowercase
+PAL_ARABIC   = " \u0627\u0628\u062a\u062b\u062c\u062d\u062e\u062f\u0630\u0631\u0632\u0633\u0634\u0635\u0636\u0637"       # Arabic letters (isolated forms)
+```
+
+#### Dot / Point Progressions
+```python
+PAL_DOTS     = " \u22c5\u2218\u2219\u25cf\u25c9\u25ce\u25c6\u2726\u2605"                   # dot size progression
+PAL_BRAILLE  = " \u2801\u2802\u2803\u2804\u2805\u2806\u2807\u2808\u2809\u280a\u280b\u280c\u280d\u280e\u280f\u2810\u2811\u2812\u2813\u2814\u2815\u2816\u2817\u2818\u2819\u281a\u281b\u281c\u281d\u281e\u281f\u283f"  # braille patterns
+PAL_STARS    = " \u00b7\u2727\u2726\u2729\u2728\u2605\u2736\u2733\u2738"               # star progression
+```
+
+#### Project-Specific (examples -- invent new ones per project)
+```python
+PAL_HERMES   = " .\u00b7~=\u2248\u221e\u26a1\u263f\u2726\u2605\u2295\u25ca\u25c6\u25b2\u25bc\u25cf\u25a0"   # mythology/tech blend
+PAL_OCEAN    = " ~\u2248\u2248\u2248\u223c\u2307\u2248\u224b\u224c\u2248"                       # water/wave characters
+PAL_ORGANIC  = " .\u00b0\u2218\u2022\u25e6\u25c9\u2742\u273f\u2741\u2743"                 # growing/botanical
+PAL_MACHINE  = " _\u2500\u2502\u250c\u2510\u253c\u2261\u25a0\u2588\u2593\u2592\u2591"             # mechanical/industrial
+```
+
+### Creating Custom Palettes
+
+When designing for a project, build palettes from the content's theme:
+
+1. **Choose a visual family** (dots, blocks, symbols, script)
+2. **Sort by visual weight** -- render each char at target font size, count lit pixels, sort ascending
+3. **Test at target grid size** -- some chars collapse to blobs at small sizes
+4. **Validate in font** -- remove chars the font can't render:
+
+```python
+def validate_palette(pal, font):
+    """Remove characters the font can't render."""
+    valid = []
+    for c in pal:
+        if c == " ":
+            valid.append(c)
+            continue
+        img = Image.new("L", (20, 20), 0)
+        ImageDraw.Draw(img).text((0, 0), c, fill=255, font=font)
+        if np.array(img).max() > 0:  # char actually rendered something
+            valid.append(c)
+    return "".join(valid)
+```
+
+### Mapping Values to Characters
+
+```python
+def val2char(v, mask, pal=PAL_DEFAULT):
+    """Map float array (0-1) to character array using palette."""
+    n = len(pal)
+    idx = np.clip((v * n).astype(int), 0, n - 1)
+    out = np.full(v.shape, " ", dtype="U1")
+    for i, ch in enumerate(pal):
+        out[mask & (idx == i)] = ch
+    return out
+```
+
+**Nonlinear mapping** for different visual curves:
+
+```python
+def val2char_gamma(v, mask, pal, gamma=1.0):
+    """Gamma-corrected palette mapping. gamma<1 = brighter, gamma>1 = darker."""
+    v_adj = np.power(np.clip(v, 0, 1), gamma)
+    return val2char(v_adj, mask, pal)
+
+def val2char_step(v, mask, pal, thresholds):
+    """Custom threshold mapping. thresholds = list of float breakpoints."""
+    out = np.full(v.shape, pal[0], dtype="U1")
+    for i, thr in enumerate(thresholds):
+        out[mask & (v > thr)] = pal[min(i + 1, len(pal) - 1)]
+    return out
+```
+
+---
+
+## Color System
+
+### HSV->RGB (Vectorized)
+
+All color computation in HSV for intuitive control, converted at render time:
+
+```python
+def hsv2rgb(h, s, v):
+    """Vectorized HSV->RGB. h,s,v are numpy arrays. Returns (R,G,B) uint8 arrays."""
+    h = h % 1.0
+    c = v * s; x = c * (1 - np.abs((h*6) % 2 - 1)); m = v - c
+    # ... 6 sector assignment ...
+    return (np.clip((r+m)*255, 0, 255).astype(np.uint8),
+            np.clip((g+m)*255, 0, 255).astype(np.uint8),
+            np.clip((b+m)*255, 0, 255).astype(np.uint8))
+```
+
+### Color Mapping Strategies
+
+Don't default to a single strategy. Choose based on the visual intent:
+
+| Strategy | Hue source | Effect | Good for |
+|----------|------------|--------|----------|
+| Angle-mapped | `g.angle / (2*pi)` | Rainbow around center | Radial effects, kaleidoscopes |
+| Distance-mapped | `g.dist_n * 0.3` | Gradient from center | Tunnels, depth effects |
+| Frequency-mapped | `f["cent"] * 0.2` | Timbral color shifting | Audio-reactive |
+| Value-mapped | `val * 0.15` | Brightness-dependent hue | Fire, heat maps |
+| Time-cycled | `t * rate` | Slow color rotation | Ambient, chill |
+| Source-sampled | Video frame pixel colors | Preserve original color | Video-to-ASCII |
+| Palette-indexed | Discrete color lookup | Flat graphic style | Retro, pixel art |
+| Temperature | Blend between warm/cool | Emotional tone | Mood-driven scenes |
+| Complementary | `hue` and `hue + 0.5` | High contrast | Bold, dramatic |
+| Triadic | `hue`, `hue + 0.33`, `hue + 0.66` | Vibrant, balanced | Psychedelic |
+| Analogous | `hue +/- 0.08` | Harmonious, subtle | Elegant, cohesive |
+| Monochrome | Fixed hue, vary S and V | Restrained, focused | Noir, minimal |
+
+### Color Palettes (Discrete RGB)
+
+For non-HSV workflows -- direct RGB color sets for graphic/retro looks:
+
+```python
+# Named color palettes -- use for flat/graphic styles or per-character coloring
+COLORS_NEON = [(255,0,102), (0,255,153), (102,0,255), (255,255,0), (0,204,255)]
+COLORS_PASTEL = [(255,179,186), (255,223,186), (255,255,186), (186,255,201), (186,225,255)]
+COLORS_MONO_GREEN = [(0,40,0), (0,80,0), (0,140,0), (0,200,0), (0,255,0)]
+COLORS_MONO_AMBER = [(40,20,0), (80,50,0), (140,90,0), (200,140,0), (255,191,0)]
+COLORS_CYBERPUNK = [(255,0,60), (0,255,200), (180,0,255), (255,200,0)]
+COLORS_VAPORWAVE = [(255,113,206), (1,205,254), (185,103,255), (5,255,161)]
+COLORS_EARTH = [(86,58,26), (139,90,43), (189,154,91), (222,193,136), (245,230,193)]
+COLORS_ICE = [(200,230,255), (150,200,240), (100,170,230), (60,130,210), (30,80,180)]
+COLORS_BLOOD = [(80,0,0), (140,10,10), (200,20,20), (255,50,30), (255,100,80)]
+COLORS_FOREST = [(10,30,10), (20,60,15), (30,100,20), (50,150,30), (80,200,50)]
+
+def rgb_palette_map(val, mask, palette):
+    """Map float array (0-1) to RGB colors from a discrete palette."""
+    n = len(palette)
+    idx = np.clip((val * n).astype(int), 0, n - 1)
+    R = np.zeros(val.shape, dtype=np.uint8)
+    G = np.zeros(val.shape, dtype=np.uint8)
+    B = np.zeros(val.shape, dtype=np.uint8)
+    for i, (r, g, b) in enumerate(palette):
+        m = mask & (idx == i)
+        R[m] = r; G[m] = g; B[m] = b
+    return R, G, B
+```
+
+### Compositing Helpers
+
+```python
+def mkc(R, G, B, rows, cols):
+    """Pack 3 uint8 arrays into (rows, cols, 3) color array."""
+    o = np.zeros((rows, cols, 3), dtype=np.uint8)
+    o[:,:,0] = R; o[:,:,1] = G; o[:,:,2] = B
+    return o
+
+def layer_over(base_ch, base_co, top_ch, top_co):
+    """Composite top layer onto base. Non-space chars overwrite."""
+    m = top_ch != " "
+    base_ch[m] = top_ch[m]; base_co[m] = top_co[m]
+    return base_ch, base_co
+
+def layer_blend(base_co, top_co, alpha):
+    """Alpha-blend top color layer onto base. alpha is float array (0-1) or scalar."""
+    if isinstance(alpha, (int, float)):
+        alpha = np.full(base_co.shape[:2], alpha, dtype=np.float32)
+    a = alpha[:,:,None]
+    return np.clip(base_co * (1 - a) + top_co * a, 0, 255).astype(np.uint8)
+
+def stamp(ch, co, text, row, col, color=(255,255,255)):
+    """Write text string at position."""
+    for i, c in enumerate(text):
+        cc = col + i
+        if 0 <= row < ch.shape[0] and 0 <= cc < ch.shape[1]:
+            ch[row, cc] = c; co[row, cc] = color
+```
+
+---
+
+## Section System
+
+Map time ranges to effect functions + shader configs + grid sizes:
+
+```python
+SECTIONS = [
+    (0.0, "void"), (3.94, "starfield"), (21.0, "matrix"),
+    (46.0, "drop"), (130.0, "glitch"), (187.0, "outro"),
+]
+
+FX_DISPATCH = {"void": fx_void, "starfield": fx_starfield, ...}
+SECTION_FX = {"void": {"vignette": 0.3, "bloom": 170}, ...}
+SECTION_GRID = {"void": "md", "starfield": "sm", "drop": "lg", ...}
+SECTION_MIRROR = {"drop": "h", "bass_rings": "quad"}
+
+def get_section(t):
+    sec = SECTIONS[0][1]
+    for ts, name in SECTIONS:
+        if t >= ts: sec = name
+    return sec
+```
+
+---
+
+## Parallel Encoding
+
+Split frames across N workers. Each pipes raw RGB to its own ffmpeg subprocess:
+
+```python
+def render_batch(batch_id, frame_start, frame_end, features, seg_path):
+    r = Renderer()
+    cmd = ["ffmpeg", "-y", "-f", "rawvideo", "-pix_fmt", "rgb24",
+           "-s", f"{VW}x{VH}", "-r", str(FPS), "-i", "pipe:0",
+           "-c:v", "libx264", "-preset", "fast", "-crf", "18",
+           "-pix_fmt", "yuv420p", seg_path]
+
+    # CRITICAL: stderr to file, not pipe
+    stderr_fh = open(os.path.join(workdir, f"err_{batch_id:02d}.log"), "w")
+    pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE,
+                            stdout=subprocess.DEVNULL, stderr=stderr_fh)
+
+    for fi in range(frame_start, frame_end):
+        t = fi / FPS
+        sec = get_section(t)
+        f = {k: float(features[k][fi]) for k in features}
+        ch, co = FX_DISPATCH[sec](r, f, t)
+        canvas = r.render(ch, co)
+        canvas = apply_mirror(canvas, sec, f)
+        canvas = apply_shaders(canvas, sec, f, t)
+        pipe.stdin.write(canvas.tobytes())
+
+    pipe.stdin.close()
+    pipe.wait()
+    stderr_fh.close()
+```
+
+Concatenate segments + mux audio:
+
+```python
+# Write concat file
+with open(concat_path, "w") as cf:
+    for seg in segments:
+        cf.write(f"file '{seg}'\n")
+
+subprocess.run(["ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", concat_path,
+                "-i", audio_path, "-c:v", "copy", "-c:a", "aac", "-b:a", "192k",
+                "-shortest", output_path])
+```
+
+## Effect Function Contract
+
+### v2 Protocol (Current)
+
+Every scene function: `(renderer, features_dict, time_float, state_dict) -> canvas_uint8`
+
+```python
+def fx_example(r, f, t, S):
+    """Scene function returns a full pixel canvas (uint8 H,W,3).
+    Scenes have full control over multi-grid rendering and pixel-level composition.
+    """
+    # Render multiple layers at different grid densities
+    canvas_a = _render_vf(r, "md", vf_plasma, hf_angle(0.0), PAL_DENSE, f, t, S)
+    canvas_b = _render_vf(r, "sm", vf_vortex, hf_time_cycle(0.1), PAL_RUNE, f, t, S)
+
+    # Pixel-level blend
+    result = blend_canvas(canvas_a, canvas_b, "screen", 0.8)
+    return result
+```
+
+See `references/scenes.md` for the full scene protocol, the Renderer class, `_render_vf()` helper, and complete scene examples.
+
+See `references/composition.md` for blend modes, tone mapping, feedback buffers, and multi-grid composition.
+
+### v1 Protocol (Legacy)
+
+Simple scenes that use a single grid can still return `(chars, colors)` and let the caller handle rendering, but the v2 canvas protocol is preferred for all new code.
+
+```python
+def fx_simple(r, f, t, S):
+    g = r.get_grid("md")
+    val = np.sin(g.dist * 0.1 - t * 3) * f.get("bass", 0.3) * 2
+    val = np.clip(val, 0, 1); mask = val > 0.03
+    ch = val2char(val, mask, PAL_DEFAULT)
+    R, G, B = hsv2rgb(np.full_like(val, 0.6), np.full_like(val, 0.7), val)
+    co = mkc(R, G, B, g.rows, g.cols)
+    return g.render(ch, co)  # returns canvas directly
+```
+
+### Persistent State
+
+Effects that need state across frames (particles, rain columns) use the `S` dict parameter (which is `r.S` — same object, but passed explicitly for clarity):
+
+```python
+def fx_with_state(r, f, t, S):
+    if "particles" not in S:
+        S["particles"] = initialize_particles()
+    update_particles(S["particles"])
+    # ...
+```
+
+State persists across frames within a single scene/clip. Each worker process (and each scene) gets its own independent state.
+
+### Helper Functions
+
+```python
+def hsv2rgb_scalar(h, s, v):
+    """Single-value HSV to RGB. Returns (R, G, B) tuple of ints 0-255."""
+    h = h % 1.0
+    c = v * s; x = c * (1 - abs((h * 6) % 2 - 1)); m = v - c
+    if h * 6 < 1:   r, g, b = c, x, 0
+    elif h * 6 < 2:  r, g, b = x, c, 0
+    elif h * 6 < 3:  r, g, b = 0, c, x
+    elif h * 6 < 4:  r, g, b = 0, x, c
+    elif h * 6 < 5:  r, g, b = x, 0, c
+    else:             r, g, b = c, 0, x
+    return (int((r+m)*255), int((g+m)*255), int((b+m)*255))
+
+def log(msg):
+    """Print timestamped log message."""
+    print(msg, flush=True)
+```
--- a/skills/creative/ascii-video/references/composition.md
+++ b/skills/creative/ascii-video/references/composition.md
@@ -0,0 +1,476 @@
+# Composition & Brightness Reference
+
+The composable system is the core of visual complexity. It operates at three levels: pixel-level blend modes, multi-grid composition, and adaptive brightness management. This document covers all three.
+
+## Pixel-Level Blend Modes
+
+### The `blend_canvas()` Function
+
+All blending operates on full pixel canvases (`uint8 H,W,3`). Internally converts to float32 [0,1] for precision, blends, lerps by opacity, converts back.
+
+```python
+def blend_canvas(base, top, mode="normal", opacity=1.0):
+    af = base.astype(np.float32) / 255.0
+    bf = top.astype(np.float32) / 255.0
+    fn = BLEND_MODES.get(mode, BLEND_MODES["normal"])
+    result = fn(af, bf)
+    if opacity < 1.0:
+        result = af * (1 - opacity) + result * opacity
+    return np.clip(result * 255, 0, 255).astype(np.uint8)
+```
+
+### 20 Blend Modes
+
+```python
+BLEND_MODES = {
+    # Basic arithmetic
+    "normal":       lambda a, b: b,
+    "add":          lambda a, b: np.clip(a + b, 0, 1),
+    "subtract":     lambda a, b: np.clip(a - b, 0, 1),
+    "multiply":     lambda a, b: a * b,
+    "screen":       lambda a, b: 1 - (1 - a) * (1 - b),
+
+    # Contrast
+    "overlay":      lambda a, b: np.where(a < 0.5, 2*a*b, 1 - 2*(1-a)*(1-b)),
+    "softlight":    lambda a, b: (1 - 2*b)*a*a + 2*b*a,
+    "hardlight":    lambda a, b: np.where(b < 0.5, 2*a*b, 1 - 2*(1-a)*(1-b)),
+
+    # Difference
+    "difference":   lambda a, b: np.abs(a - b),
+    "exclusion":    lambda a, b: a + b - 2*a*b,
+
+    # Dodge / burn
+    "colordodge":   lambda a, b: np.clip(a / (1 - b + 1e-6), 0, 1),
+    "colorburn":    lambda a, b: np.clip(1 - (1 - a) / (b + 1e-6), 0, 1),
+
+    # Light
+    "linearlight":  lambda a, b: np.clip(a + 2*b - 1, 0, 1),
+    "vividlight":   lambda a, b: np.where(b < 0.5,
+                        np.clip(1 - (1-a)/(2*b + 1e-6), 0, 1),
+                        np.clip(a / (2*(1-b) + 1e-6), 0, 1)),
+    "pin_light":    lambda a, b: np.where(b < 0.5,
+                        np.minimum(a, 2*b), np.maximum(a, 2*b - 1)),
+    "hard_mix":     lambda a, b: np.where(a + b >= 1.0, 1.0, 0.0),
+
+    # Compare
+    "lighten":      lambda a, b: np.maximum(a, b),
+    "darken":       lambda a, b: np.minimum(a, b),
+
+    # Grain
+    "grain_extract": lambda a, b: np.clip(a - b + 0.5, 0, 1),
+    "grain_merge":  lambda a, b: np.clip(a + b - 0.5, 0, 1),
+}
+```
+
+### Blend Mode Selection Guide
+
+**Modes that brighten** (safe for dark inputs):
+- `screen` — always brightens. Two 50% gray layers screen to 75%. The go-to safe blend.
+- `add` — simple addition, clips at white. Good for sparkles, glows, particle overlays.
+- `colordodge` — extreme brightening at overlap zones. Can blow out. Use low opacity (0.3-0.5).
+- `linearlight` — aggressive brightening. Similar to add but with offset.
+
+**Modes that darken** (avoid with dark inputs):
+- `multiply` — darkens everything. Only use when both layers are already bright.
+- `overlay` — darkens when base < 0.5, brightens when base > 0.5. Crushes dark inputs: `2 * 0.12 * 0.12 = 0.03`. Use `screen` instead for dark material.
+- `colorburn` — extreme darkening at overlap zones.
+
+**Modes that create contrast**:
+- `softlight` — gentle contrast. Good for subtle texture overlay.
+- `hardlight` — strong contrast. Like overlay but keyed on the top layer.
+- `vividlight` — very aggressive contrast. Use sparingly.
+
+**Modes that create color effects**:
+- `difference` — XOR-like patterns. Two identical layers difference to black; offset layers create wild colors. Great for psychedelic looks.
+- `exclusion` — softer version of difference. Creates complementary color patterns.
+- `hard_mix` — posterizes to pure black/white/saturated color at intersections.
+
+**Modes for texture blending**:
+- `grain_extract` / `grain_merge` — extract a texture from one layer, apply it to another.
+
+### Multi-Layer Chaining
+
+```python
+# Pattern: render layers -> blend sequentially
+canvas_a = _render_vf(r, "md", vf_plasma, hf_angle(0.0), PAL_DENSE, f, t, S)
+canvas_b = _render_vf(r, "sm", vf_vortex, hf_time_cycle(0.1), PAL_RUNE, f, t, S)
+canvas_c = _render_vf(r, "lg", vf_rings, hf_distance(), PAL_BLOCKS, f, t, S)
+
+result = blend_canvas(canvas_a, canvas_b, "screen", 0.8)
+result = blend_canvas(result, canvas_c, "difference", 0.6)
+```
+
+Order matters: `screen(A, B)` is commutative, but `difference(screen(A,B), C)` differs from `difference(A, screen(B,C))`.
+
+---
+
+## Multi-Grid Composition
+
+This is the core visual technique. Rendering the same conceptual scene at different grid densities (character sizes) creates natural texture interference, because characters at different scales overlap at different spatial frequencies.
+
+### Why It Works
+
+- `sm` grid (10pt font): 320x83 characters. Fine detail, dense texture.
+- `md` grid (16pt): 192x56 characters. Medium density.
+- `lg` grid (20pt): 160x45 characters. Coarse, chunky characters.
+
+When you render a plasma field on `sm` and a vortex on `lg`, then screen-blend them, the fine plasma texture shows through the gaps in the coarse vortex characters. The result has more visual complexity than either layer alone.
+
+### The `_render_vf()` Helper
+
+This is the workhorse function. It takes a value field + hue field + palette + grid, renders to a complete pixel canvas:
+
+```python
+def _render_vf(r, grid_key, val_fn, hue_fn, pal, f, t, S, sat=0.8, threshold=0.03):
+    """Render a value field + hue field to a pixel canvas via a named grid.
+
+    Args:
+        r: Renderer instance (has .get_grid())
+        grid_key: "xs", "sm", "md", "lg", "xl", "xxl"
+        val_fn: (g, f, t, S) -> float32 [0,1] array (rows, cols)
+        hue_fn: callable (g, f, t, S) -> float32 hue array, OR float scalar
+        pal: character palette string
+        f: feature dict
+        t: time in seconds
+        S: persistent state dict
+        sat: HSV saturation (0-1)
+        threshold: minimum value to render (below = space)
+
+    Returns:
+        uint8 array (VH, VW, 3) — full pixel canvas
+    """
+    g = r.get_grid(grid_key)
+    val = np.clip(val_fn(g, f, t, S), 0, 1)
+    mask = val > threshold
+    ch = val2char(val, mask, pal)
+
+    # Hue: either a callable or a fixed float
+    if callable(hue_fn):
+        h = hue_fn(g, f, t, S) % 1.0
+    else:
+        h = np.full((g.rows, g.cols), float(hue_fn), dtype=np.float32)
+
+    # CRITICAL: broadcast to full shape and copy (see Troubleshooting)
+    h = np.broadcast_to(h, (g.rows, g.cols)).copy()
+
+    R, G, B = hsv2rgb(h, np.full_like(val, sat), val)
+    co = mkc(R, G, B, g.rows, g.cols)
+    return g.render(ch, co)
+```
+
+### Grid Combination Strategies
+
+| Combination | Effect | Good For |
+|-------------|--------|----------|
+| `sm` + `lg` | Maximum contrast between fine detail and chunky blocks | Bold, graphic looks |
+| `sm` + `md` | Subtle texture layering, similar scales | Organic, flowing looks |
+| `md` + `lg` + `xs` | Three-scale interference, maximum complexity | Psychedelic, dense |
+| `sm` + `sm` (different effects) | Same scale, pattern interference only | Moire, interference |
+
+### Complete Multi-Grid Scene Example
+
+```python
+def fx_psychedelic(r, f, t, S):
+    """Three-layer multi-grid scene with beat-reactive kaleidoscope."""
+    # Layer A: plasma on medium grid with rainbow hue
+    canvas_a = _render_vf(r, "md",
+        lambda g, f, t, S: vf_plasma(g, f, t, S) * 1.3,
+        hf_angle(0.0), PAL_DENSE, f, t, S, sat=0.8)
+
+    # Layer B: vortex on small grid with cycling hue
+    canvas_b = _render_vf(r, "sm",
+        lambda g, f, t, S: vf_vortex(g, f, t, S, twist=5.0) * 1.2,
+        hf_time_cycle(0.1), PAL_RUNE, f, t, S, sat=0.7)
+
+    # Layer C: rings on large grid with distance hue
+    canvas_c = _render_vf(r, "lg",
+        lambda g, f, t, S: vf_rings(g, f, t, S, n_base=8, spacing_base=3) * 1.4,
+        hf_distance(0.3, 0.02), PAL_BLOCKS, f, t, S, sat=0.9)
+
+    # Blend: A screened with B, then difference with C
+    result = blend_canvas(canvas_a, canvas_b, "screen", 0.8)
+    result = blend_canvas(result, canvas_c, "difference", 0.6)
+
+    # Beat-triggered kaleidoscope
+    if f.get("bdecay", 0) > 0.3:
+        result = sh_kaleidoscope(result.copy(), folds=6)
+
+    return result
+```
+
+---
+
+## Adaptive Tone Mapping
+
+### The Brightness Problem
+
+ASCII characters are small bright dots on a black background. Most pixels in any frame are background (black). This means:
+- Mean frame brightness is inherently low (often 5-30 out of 255)
+- Different effect combinations produce wildly different brightness levels
+- A spiral scene might be 50 mean, while a fire scene is 9 mean
+- Linear multipliers (e.g., `canvas * 2.0`) either leave dark scenes dark or blow out bright scenes
+
+### The `tonemap()` Function
+
+Replaces linear brightness multipliers with adaptive per-frame normalization + gamma correction:
+
+```python
+def tonemap(canvas, target_mean=90, gamma=0.75, black_point=2, white_point=253):
+    """Adaptive tone-mapping: normalizes + gamma-corrects so no frame is
+    fully dark or washed out.
+
+    1. Compute 1st and 99.5th percentile (ignores outlier pixels)
+    2. Stretch that range to [0, 1]
+    3. Apply gamma curve (< 1 lifts shadows, > 1 darkens)
+    4. Rescale to [black_point, white_point]
+    """
+    f = canvas.astype(np.float32)
+    lo = np.percentile(f, 1)
+    hi = np.percentile(f, 99.5)
+    if hi - lo < 10:
+        hi = max(hi, lo + 10)  # near-uniform frame fallback
+    f = np.clip((f - lo) / (hi - lo), 0.0, 1.0)
+    f = np.power(f, gamma)
+    f = f * (white_point - black_point) + black_point
+    return np.clip(f, 0, 255).astype(np.uint8)
+```
+
+### Why Gamma, Not Linear
+
+Linear multiplier `* 2.0`:
+```
+input 10  -> output 20   (still dark)
+input 100 -> output 200  (ok)
+input 200 -> output 255  (clipped, lost detail)
+```
+
+Gamma 0.75 after normalization:
+```
+input 0.04 -> output 0.08 (lifted from invisible to visible)
+input 0.39 -> output 0.50 (moderate lift)
+input 0.78 -> output 0.84 (gentle lift, no clipping)
+```
+
+Gamma < 1 compresses the highlights and expands the shadows. This is exactly what we need: lift dark ASCII content into visibility without blowing out the bright parts.
+
+### Pipeline Ordering
+
+The pipeline in `render_clip()` is:
+
+```
+scene_fn(r, f, t, S)  ->  canvas
+         |
+    tonemap(canvas, gamma=scene_gamma)
+         |
+    FeedbackBuffer.apply(canvas, ...)
+         |
+    ShaderChain.apply(canvas, f=f, t=t)
+         |
+    ffmpeg pipe
+```
+
+Tonemap runs BEFORE feedback and shaders. This means:
+- Feedback operates on normalized data (consistent behavior regardless of scene brightness)
+- Shaders like solarize, posterize, contrast operate on properly-ranged data
+- The brightness shader in the chain is no longer needed (tonemap handles it)
+
+### Per-Scene Gamma Tuning
+
+Default gamma is 0.75. Scenes that apply destructive post-processing need more aggressive lift because the destruction happens after tonemap:
+
+| Scene Type | Recommended Gamma | Why |
+|------------|-------------------|-----|
+| Standard effects | 0.75 | Default, works for most scenes |
+| Solarize post-process | 0.50-0.60 | Solarize inverts bright pixels, reducing overall brightness |
+| Posterize post-process | 0.50-0.55 | Posterize quantizes, often crushing mid-values to black |
+| Heavy difference blending | 0.60-0.70 | Difference mode creates many near-zero pixels |
+| Already bright scenes | 0.85-1.0 | Don't over-boost scenes that are naturally bright |
+
+Configure via the scene table:
+
+```python
+SCENES = [
+    {"start": 9.17, "end": 11.25, "name": "fire", "gamma": 0.55,
+     "fx": fx_fire, "shaders": [("solarize", {"threshold": 200}), ...]},
+    {"start": 25.96, "end": 27.29, "name": "diamond", "gamma": 0.5,
+     "fx": fx_diamond, "shaders": [("bloom", {"thr": 90}), ...]},
+]
+```
+
+### Brightness Verification
+
+After rendering, spot-check frame brightness:
+
+```python
+# In test-frame mode
+canvas = scene["fx"](r, feat, t, r.S)
+canvas = tonemap(canvas, gamma=scene.get("gamma", 0.75))
+chain = ShaderChain()
+for sn, kw in scene.get("shaders", []):
+    chain.add(sn, **kw)
+canvas = chain.apply(canvas, f=feat, t=t)
+print(f"Mean brightness: {canvas.astype(float).mean():.1f}, max: {canvas.max()}")
+```
+
+Target ranges after tonemap + shaders:
+- Quiet/ambient scenes: mean 30-60
+- Active scenes: mean 40-100
+- Climax/peak scenes: mean 60-150
+- If mean < 20: gamma is too high or a shader is destroying brightness
+- If mean > 180: gamma is too low or add is stacking too much
+
+---
+
+## FeedbackBuffer Spatial Transforms
+
+The feedback buffer stores the previous frame and blends it into the current frame with decay. Spatial transforms applied to the buffer before blending create the illusion of motion in the feedback trail.
+
+### Implementation
+
+```python
+class FeedbackBuffer:
+    def __init__(self):
+        self.buf = None
+
+    def apply(self, canvas, decay=0.85, blend="screen", opacity=0.5,
+              transform=None, transform_amt=0.02, hue_shift=0.0):
+        if self.buf is None:
+            self.buf = canvas.astype(np.float32) / 255.0
+            return canvas
+
+        # Decay old buffer
+        self.buf *= decay
+
+        # Spatial transform
+        if transform:
+            self.buf = self._transform(self.buf, transform, transform_amt)
+
+        # Hue shift the feedback for rainbow trails
+        if hue_shift > 0:
+            self.buf = self._hue_shift(self.buf, hue_shift)
+
+        # Blend feedback into current frame
+        result = blend_canvas(canvas,
+                              np.clip(self.buf * 255, 0, 255).astype(np.uint8),
+                              blend, opacity)
+
+        # Update buffer with current frame
+        self.buf = result.astype(np.float32) / 255.0
+        return result
+
+    def _transform(self, buf, transform, amt):
+        h, w = buf.shape[:2]
+        if transform == "zoom":
+            # Zoom in: sample from slightly inside (creates expanding tunnel)
+            m = int(h * amt); n = int(w * amt)
+            if m > 0 and n > 0:
+                cropped = buf[m:-m or None, n:-n or None]
+                # Resize back to full (nearest-neighbor for speed)
+                buf = np.array(Image.fromarray(
+                    np.clip(cropped * 255, 0, 255).astype(np.uint8)
+                ).resize((w, h), Image.NEAREST)).astype(np.float32) / 255.0
+        elif transform == "shrink":
+            # Zoom out: pad edges, shrink center
+            m = int(h * amt); n = int(w * amt)
+            small = np.array(Image.fromarray(
+                np.clip(buf * 255, 0, 255).astype(np.uint8)
+            ).resize((w - 2*n, h - 2*m), Image.NEAREST))
+            new = np.zeros((h, w, 3), dtype=np.uint8)
+            new[m:m+small.shape[0], n:n+small.shape[1]] = small
+            buf = new.astype(np.float32) / 255.0
+        elif transform == "rotate_cw":
+            # Small clockwise rotation via affine
+            angle = amt * 10  # amt=0.005 -> 0.05 degrees per frame
+            cy, cx = h / 2, w / 2
+            Y = np.arange(h, dtype=np.float32)[:, None]
+            X = np.arange(w, dtype=np.float32)[None, :]
+            cos_a, sin_a = np.cos(angle), np.sin(angle)
+            sx = (X - cx) * cos_a + (Y - cy) * sin_a + cx
+            sy = -(X - cx) * sin_a + (Y - cy) * cos_a + cy
+            sx = np.clip(sx.astype(int), 0, w - 1)
+            sy = np.clip(sy.astype(int), 0, h - 1)
+            buf = buf[sy, sx]
+        elif transform == "rotate_ccw":
+            angle = -amt * 10
+            cy, cx = h / 2, w / 2
+            Y = np.arange(h, dtype=np.float32)[:, None]
+            X = np.arange(w, dtype=np.float32)[None, :]
+            cos_a, sin_a = np.cos(angle), np.sin(angle)
+            sx = (X - cx) * cos_a + (Y - cy) * sin_a + cx
+            sy = -(X - cx) * sin_a + (Y - cy) * cos_a + cy
+            sx = np.clip(sx.astype(int), 0, w - 1)
+            sy = np.clip(sy.astype(int), 0, h - 1)
+            buf = buf[sy, sx]
+        elif transform == "shift_up":
+            pixels = max(1, int(h * amt))
+            buf = np.roll(buf, -pixels, axis=0)
+            buf[-pixels:] = 0  # black fill at bottom
+        elif transform == "shift_down":
+            pixels = max(1, int(h * amt))
+            buf = np.roll(buf, pixels, axis=0)
+            buf[:pixels] = 0
+        elif transform == "mirror_h":
+            buf = buf[:, ::-1]
+        return buf
+
+    def _hue_shift(self, buf, amount):
+        """Rotate hues of the feedback buffer. Operates on float32 [0,1]."""
+        rgb = np.clip(buf * 255, 0, 255).astype(np.uint8)
+        hsv = np.zeros_like(buf)
+        # Simple approximate RGB->HSV->shift->RGB
+        r, g, b = buf[:,:,0], buf[:,:,1], buf[:,:,2]
+        mx = np.maximum(np.maximum(r, g), b)
+        mn = np.minimum(np.minimum(r, g), b)
+        delta = mx - mn + 1e-10
+        # Hue
+        h = np.where(mx == r, ((g - b) / delta) % 6,
+            np.where(mx == g, (b - r) / delta + 2, (r - g) / delta + 4))
+        h = (h / 6 + amount) % 1.0
+        # Reconstruct with shifted hue (simplified)
+        s = delta / (mx + 1e-10)
+        v = mx
+        c = v * s; x = c * (1 - np.abs((h * 6) % 2 - 1)); m = v - c
+        ro = np.zeros_like(h); go = np.zeros_like(h); bo = np.zeros_like(h)
+        for lo, hi, rv, gv, bv in [(0,1,c,x,0),(1,2,x,c,0),(2,3,0,c,x),
+                                     (3,4,0,x,c),(4,5,x,0,c),(5,6,c,0,x)]:
+            mask = ((h*6) >= lo) & ((h*6) < hi)
+            ro[mask] = rv[mask] if not isinstance(rv, (int,float)) else rv
+            go[mask] = gv[mask] if not isinstance(gv, (int,float)) else gv
+            bo[mask] = bv[mask] if not isinstance(bv, (int,float)) else bv
+        return np.stack([ro+m, go+m, bo+m], axis=2)
+```
+
+### Feedback Presets
+
+| Preset | Config | Visual Effect |
+|--------|--------|---------------|
+| Infinite zoom tunnel | `decay=0.8, blend="screen", transform="zoom", transform_amt=0.015` | Expanding ring patterns |
+| Rainbow trails | `decay=0.7, blend="screen", transform="zoom", transform_amt=0.01, hue_shift=0.02` | Psychedelic color trails |
+| Ghostly echo | `decay=0.9, blend="add", opacity=0.15, transform="shift_up", transform_amt=0.01` | Faint upward smearing |
+| Kaleidoscopic recursion | `decay=0.75, blend="screen", transform="rotate_cw", transform_amt=0.005, hue_shift=0.01` | Rotating mandala feedback |
+| Color evolution | `decay=0.8, blend="difference", opacity=0.4, hue_shift=0.03` | Frame-to-frame color XOR |
+| Rising heat haze | `decay=0.5, blend="add", opacity=0.2, transform="shift_up", transform_amt=0.02` | Hot air shimmer |
+
+---
+
+## PixelBlendStack
+
+Higher-level wrapper for multi-layer compositing:
+
+```python
+class PixelBlendStack:
+    def __init__(self):
+        self.layers = []
+
+    def add(self, canvas, mode="normal", opacity=1.0):
+        self.layers.append((canvas, mode, opacity))
+        return self
+
+    def composite(self):
+        if not self.layers:
+            return np.zeros((VH, VW, 3), dtype=np.uint8)
+        result = self.layers[0][0]
+        for canvas, mode, opacity in self.layers[1:]:
+            result = blend_canvas(result, canvas, mode, opacity)
+        return result
+```
--- a/skills/creative/ascii-video/references/effects.md
+++ b/skills/creative/ascii-video/references/effects.md
@@ -0,0 +1,893 @@
+# Effect Catalog
+
+Effect building blocks that produce visual patterns. In v2, these are used **inside scene functions** that return a pixel canvas directly. The building blocks below operate on grid coordinate arrays and produce `(chars, colors)` or value/hue fields that the scene function renders to canvas via `_render_vf()`. See `composition.md` for the v2 rendering pattern and `scenes.md` for scene function examples.
+
+## Design Philosophy
+
+Effects are the creative core. Don't copy these verbatim for every project -- use them as **building blocks** and **combine, modify, and invent** new ones. Every project should feel distinct.
+
+Key principles:
+- **Layer multiple effects** rather than using a single monolithic function
+- **Parameterize everything** -- hue, speed, density, amplitude should all be arguments
+- **React to features** -- audio/video features should modulate at least 2-3 parameters per effect
+- **Vary per section** -- never use the same effect config for the entire video
+- **Invent project-specific effects** -- the catalog below is a starting vocabulary, not a fixed set
+
+---
+
+## Background Fills
+
+Every effect should start with a background. Never leave flat black.
+
+### Animated Sine Field (General Purpose)
+```python
+def bg_sinefield(g, f, t, hue=0.6, bri=0.5, pal=PAL_DEFAULT,
+                 freq=(0.13, 0.17, 0.07, 0.09), speed=(0.5, -0.4, -0.3, 0.2)):
+    """Layered sine field. Adjust freq/speed tuples for different textures."""
+    v1 = np.sin(g.cc*freq[0] + t*speed[0]) * np.sin(g.rr*freq[1] - t*speed[1]) * 0.5 + 0.5
+    v2 = np.sin(g.cc*freq[2] - t*speed[2] + g.rr*freq[3]) * 0.4 + 0.5
+    v3 = np.sin(g.dist_n*5 + t*0.2) * 0.3 + 0.4
+    v4 = np.cos(g.angle*3 - t*0.6) * 0.15 + 0.5
+    val = np.clip((v1*0.3 + v2*0.25 + v3*0.25 + v4*0.2) * bri * (0.6 + f["rms"]*0.6), 0.06, 1)
+    mask = val > 0.03
+    ch = val2char(val, mask, pal)
+    h = np.full_like(val, hue) + f.get("cent", 0.5)*0.1 + val*0.08
+    R, G, B = hsv2rgb(h, np.clip(0.35+f.get("flat",0.4)*0.4, 0, 1) * np.ones_like(val), val)
+    return ch, mkc(R, G, B, g.rows, g.cols)
+```
+
+### Video-Source Background
+```python
+def bg_video(g, frame_rgb, pal=PAL_DEFAULT, brightness=0.5):
+    small = np.array(Image.fromarray(frame_rgb).resize((g.cols, g.rows)))
+    lum = np.mean(small, axis=2) / 255.0 * brightness
+    mask = lum > 0.02
+    ch = val2char(lum, mask, pal)
+    co = np.clip(small * np.clip(lum[:,:,None]*1.5+0.3, 0.3, 1), 0, 255).astype(np.uint8)
+    return ch, co
+```
+
+### Noise / Static Field
+```python
+def bg_noise(g, f, t, pal=PAL_BLOCKS, density=0.3, hue_drift=0.02):
+    val = np.random.random((g.rows, g.cols)).astype(np.float32) * density * (0.5 + f["rms"]*0.5)
+    val = np.clip(val, 0, 1); mask = val > 0.02
+    ch = val2char(val, mask, pal)
+    R, G, B = hsv2rgb(np.full_like(val, t*hue_drift % 1), np.full_like(val, 0.3), val)
+    return ch, mkc(R, G, B, g.rows, g.cols)
+```
+
+### Perlin-Like Smooth Noise
+```python
+def bg_smooth_noise(g, f, t, hue=0.5, bri=0.5, pal=PAL_DOTS, octaves=3):
+    """Layered sine approximation of Perlin noise. Cheap, smooth, organic."""
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for i in range(octaves):
+        freq = 0.05 * (2 ** i)
+        amp = 0.5 / (i + 1)
+        phase = t * (0.3 + i * 0.2)
+        val += np.sin(g.cc * freq + phase) * np.cos(g.rr * freq * 0.7 - phase * 0.5) * amp
+    val = np.clip(val * 0.5 + 0.5, 0, 1) * bri
+    mask = val > 0.03
+    ch = val2char(val, mask, pal)
+    h = np.full_like(val, hue) + val * 0.1
+    R, G, B = hsv2rgb(h, np.full_like(val, 0.5), val)
+    return ch, mkc(R, G, B, g.rows, g.cols)
+```
+
+### Cellular / Voronoi Approximation
+```python
+def bg_cellular(g, f, t, n_centers=12, hue=0.5, bri=0.6, pal=PAL_BLOCKS):
+    """Voronoi-like cells using distance to nearest of N moving centers."""
+    rng = np.random.RandomState(42)  # deterministic centers
+    cx = (rng.rand(n_centers) * g.cols).astype(np.float32)
+    cy = (rng.rand(n_centers) * g.rows).astype(np.float32)
+    # Animate centers
+    cx_t = cx + np.sin(t * 0.5 + np.arange(n_centers) * 0.7) * 5
+    cy_t = cy + np.cos(t * 0.4 + np.arange(n_centers) * 0.9) * 3
+    # Min distance to any center
+    min_d = np.full((g.rows, g.cols), 999.0, dtype=np.float32)
+    for i in range(n_centers):
+        d = np.sqrt((g.cc - cx_t[i])**2 + (g.rr - cy_t[i])**2)
+        min_d = np.minimum(min_d, d)
+    val = np.clip(1.0 - min_d / (g.cols * 0.3), 0, 1) * bri
+    # Cell edges (where distance is near-equal between two centers)
+    # ... second-nearest trick for edge highlighting
+    mask = val > 0.03
+    ch = val2char(val, mask, pal)
+    R, G, B = hsv2rgb(np.full_like(val, hue) + min_d * 0.005, np.full_like(val, 0.5), val)
+    return ch, mkc(R, G, B, g.rows, g.cols)
+```
+
+---
+
+## Radial Effects
+
+### Concentric Rings
+Bass/sub-driven pulsing rings from center. Scale ring count and thickness with bass energy.
+```python
+def eff_rings(g, f, t, hue=0.5, n_base=6, pal=PAL_DEFAULT):
+    n_rings = int(n_base + f["sub_r"] * 25 + f["bass"] * 10)
+    spacing = 2 + f["bass_r"] * 7 + f["rms"] * 3
+    ring_cv = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for ri in range(n_rings):
+        rad = (ri+1) * spacing + f["bdecay"] * 15
+        wobble = f["mid_r"]*5*np.sin(g.angle*3 + t*4) + f["hi_r"]*3*np.sin(g.angle*7 - t*6)
+        rd = np.abs(g.dist - rad - wobble)
+        th = 1 + f["sub"] * 3
+        ring_cv = np.maximum(ring_cv, np.clip((1 - rd/th) * (0.4 + f["bass"]*0.8), 0, 1))
+    # Color by angle + distance for rainbow rings
+    h = g.angle/(2*np.pi) + g.dist*0.005 + f["sub_r"]*0.2
+    return ring_cv, h
+```
+
+### Radial Rays
+```python
+def eff_rays(g, f, t, n_base=8, hue=0.5):
+    n_rays = int(n_base + f["hi_r"] * 25)
+    ray = np.clip(np.cos(g.angle*n_rays + t*3) * f["bdecay"]*0.6 * (1-g.dist_n), 0, 0.7)
+    return ray
+```
+
+### Spiral Arms (Logarithmic)
+```python
+def eff_spiral(g, f, t, n_arms=3, tightness=2.5, hue=0.5):
+    arm_cv = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for ai in range(n_arms):
+        offset = ai * 2*np.pi / n_arms
+        log_r = np.log(g.dist + 1) * tightness
+        arm_phase = g.angle + offset - log_r + t * 0.8
+        arm_val = np.clip(np.cos(arm_phase * n_arms) * 0.6 + 0.2, 0, 1)
+        arm_val *= (0.4 + f["rms"]*0.6) * np.clip(1 - g.dist_n*0.5, 0.2, 1)
+        arm_cv = np.maximum(arm_cv, arm_val)
+    return arm_cv
+```
+
+### Center Glow / Pulse
+```python
+def eff_glow(g, f, t, intensity=0.6, spread=2.0):
+    return np.clip(intensity * np.exp(-g.dist_n * spread) * (0.5 + f["rms"]*2 + np.sin(t*1.2)*0.2), 0, 0.9)
+```
+
+### Tunnel / Depth
+```python
+def eff_tunnel(g, f, t, speed=3.0, complexity=6):
+    tunnel_d = 1.0 / (g.dist_n + 0.1)
+    v1 = np.sin(tunnel_d*2 - t*speed) * 0.45 + 0.55
+    v2 = np.sin(g.angle*complexity + tunnel_d*1.5 - t*2) * 0.35 + 0.55
+    return v1 * 0.5 + v2 * 0.5
+```
+
+### Vortex (Rotating Distortion)
+```python
+def eff_vortex(g, f, t, twist=3.0, pulse=True):
+    """Twisting radial pattern -- distance modulates angle."""
+    twisted = g.angle + g.dist_n * twist * np.sin(t * 0.5)
+    val = np.sin(twisted * 4 - t * 2) * 0.5 + 0.5
+    if pulse:
+        val *= 0.5 + f.get("bass", 0.3) * 0.8
+    return np.clip(val, 0, 1)
+```
+
+---
+
+## Wave Effects
+
+### Multi-Band Frequency Waves
+Each frequency band draws its own wave at different spatial/temporal frequencies:
+```python
+def eff_freq_waves(g, f, t, bands=None):
+    if bands is None:
+        bands = [("sub",0.06,1.2,0.0), ("bass",0.10,2.0,0.08), ("lomid",0.15,3.0,0.16),
+                 ("mid",0.22,4.5,0.25), ("himid",0.32,6.5,0.4), ("hi",0.45,8.5,0.55)]
+    mid = g.rows / 2.0
+    composite = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for band_key, sf, tf, hue_base in bands:
+        amp = f.get(band_key, 0.3) * g.rows * 0.4
+        y_wave = mid - np.sin(g.cc*sf + t*tf) * amp
+        y_wave += np.sin(g.cc*sf*2.3 + t*tf*1.7) * amp * 0.2  # harmonic
+        dist = np.abs(g.rr - y_wave)
+        thickness = 2 + f.get(band_key, 0.3) * 5
+        intensity = np.clip((1 - dist/thickness) * f.get(band_key, 0.3) * 1.5, 0, 1)
+        composite = np.maximum(composite, intensity)
+    return composite
+```
+
+### Interference Pattern
+6-8 overlapping sine waves creating moire-like patterns:
+```python
+def eff_interference(g, f, t, n_waves=5):
+    """Parametric interference -- vary n_waves for complexity."""
+    # Each wave has different orientation, frequency, and feature driver
+    drivers = ["mid_r", "himid_r", "bass_r", "lomid_r", "hi_r"]
+    vals = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for i in range(min(n_waves, len(drivers))):
+        angle = i * np.pi / n_waves  # spread orientations
+        freq = 0.06 + i * 0.03
+        sp = 0.5 + i * 0.3
+        proj = g.cc * np.cos(angle) + g.rr * np.sin(angle)
+        vals += np.sin(proj * freq + t * sp) * f.get(drivers[i], 0.3) * 2.5
+    return np.clip(vals * 0.12 + 0.45, 0.1, 1)
+```
+
+### Aurora / Horizontal Bands
+```python
+def eff_aurora(g, f, t, hue=0.4, n_bands=3):
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for i in range(n_bands):
+        freq_r = 0.08 + i * 0.04
+        freq_c = 0.012 + i * 0.008
+        sp_r = 0.7 + i * 0.3
+        sp_c = 0.18 + i * 0.12
+        val += np.sin(g.rr*freq_r + t*sp_r) * np.sin(g.cc*freq_c + t*sp_c) * (0.6 / n_bands)
+    return np.clip(val * (f.get("lomid_r", 0.3)*3 + 0.2), 0, 0.7)
+```
+
+### Ripple (Point-Source Waves)
+```python
+def eff_ripple(g, f, t, sources=None, freq=0.3, damping=0.02):
+    """Concentric ripples from point sources. Sources = [(row_frac, col_frac), ...]"""
+    if sources is None:
+        sources = [(0.5, 0.5)]  # center
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for ry, rx in sources:
+        dy = g.rr - g.rows * ry
+        dx = g.cc - g.cols * rx
+        d = np.sqrt(dy**2 + dx**2)
+        val += np.sin(d * freq - t * 4) * np.exp(-d * damping) * 0.5
+    return np.clip(val + 0.5, 0, 1)
+```
+
+---
+
+## Particle Systems
+
+### General Pattern
+All particle systems use persistent state:
+```python
+S = state  # dict persisted across frames
+if "px" not in S:
+    S["px"]=[]; S["py"]=[]; S["vx"]=[]; S["vy"]=[]; S["life"]=[]; S["char"]=[]
+
+# Emit new particles (on beat, continuously, or on trigger)
+# Update: position += velocity, apply forces, decay life
+# Draw: map to grid, set char/color based on life
+# Cull: remove dead, cap total count
+```
+
+### Particle Character Sets
+
+Don't hardcode particle chars. Choose per project/mood:
+
+```python
+# Energy / explosive
+PART_ENERGY  = list("*+#@\u26a1\u2726\u2605\u2588\u2593")
+PART_SPARK   = list("\u00b7\u2022\u25cf\u2605\u2736*+")
+# Organic / natural
+PART_LEAF    = list("\u2740\u2741\u2742\u2743\u273f\u2618\u2022")
+PART_SNOW    = list("\u2744\u2745\u2746\u00b7\u2022*\u25cb")
+PART_RAIN    = list("|\u2502\u2503\u2551/\\")
+PART_BUBBLE  = list("\u25cb\u25ce\u25c9\u25cf\u2218\u2219\u00b0")
+# Data / tech
+PART_DATA    = list("01{}[]<>|/\\")
+PART_HEX     = list("0123456789ABCDEF")
+PART_BINARY  = list("01")
+# Mystical
+PART_RUNE    = list("\u16a0\u16a2\u16a6\u16b1\u16b7\u16c1\u16c7\u16d2\u16d6\u16da\u16de\u16df\u2726\u2605")
+PART_ZODIAC  = list("\u2648\u2649\u264a\u264b\u264c\u264d\u264e\u264f\u2650\u2651\u2652\u2653")
+# Minimal
+PART_DOT     = list("\u00b7\u2022\u25cf")
+PART_DASH    = list("-=~\u2500\u2550")
+```
+
+### Explosion (Beat-Triggered)
+```python
+def emit_explosion(S, f, center_r, center_c, char_set=PART_ENERGY, count_base=80):
+    if f.get("beat", 0) > 0:
+        for _ in range(int(count_base + f["rms"]*150)):
+            ang = random.uniform(0, 2*math.pi)
+            sp = random.uniform(1, 9) * (0.5 + f.get("sub_r", 0.3)*2)
+            S["px"].append(float(center_c))
+            S["py"].append(float(center_r))
+            S["vx"].append(math.cos(ang)*sp*2.5)
+            S["vy"].append(math.sin(ang)*sp)
+            S["life"].append(1.0)
+            S["char"].append(random.choice(char_set))
+# Update: gravity on vy += 0.03, life -= 0.015
+# Color: life * 255 for brightness, hue fade controlled by caller
+```
+
+### Rising Embers
+```python
+# Emit: sy = rows-1, vy = -random.uniform(1,5), vx = random.uniform(-1.5,1.5)
+# Update: vx += random jitter * 0.3, life -= 0.01
+# Cap at ~1500 particles
+```
+
+### Dissolving Cloud
+```python
+# Init: N=600 particles spread across screen
+# Update: slow upward drift, fade life progressively
+# life -= 0.002 * (1 + elapsed * 0.05)  # accelerating fade
+```
+
+### Starfield (3D Projection)
+```python
+# N stars with (sx, sy, sz) in normalized coords
+# Move: sz -= speed (stars approach camera)
+# Project: px = cx + sx/sz * cx, py = cy + sy/sz * cy
+# Reset stars that pass camera (sz <= 0.01)
+# Brightness = (1 - sz), draw streaks behind bright stars
+```
+
+### Orbit (Circular/Elliptical Motion)
+```python
+def emit_orbit(S, n=20, radius=15, speed=1.0, char_set=PART_DOT):
+    """Particles orbiting a center point."""
+    for i in range(n):
+        angle = i * 2 * math.pi / n
+        S["px"].append(0.0); S["py"].append(0.0)  # will be computed from angle
+        S["vx"].append(angle)  # store angle as "vx" for orbit
+        S["vy"].append(radius + random.uniform(-2, 2))  # store radius
+        S["life"].append(1.0)
+        S["char"].append(random.choice(char_set))
+# Update: angle += speed * dt, px = cx + radius * cos(angle), py = cy + radius * sin(angle)
+```
+
+### Gravity Well
+```python
+# Particles attracted toward one or more gravity points
+# Update: compute force vector toward each well, apply as acceleration
+# Particles that reach well center respawn at edges
+```
+
+---
+
+## Rain / Matrix Effects
+
+### Column Rain (Vectorized)
+```python
+def eff_matrix_rain(g, f, t, state, hue=0.33, bri=0.6, pal=PAL_KATA,
+                    speed_base=0.5, speed_beat=3.0):
+    """Vectorized matrix rain. state dict persists column positions."""
+    if "ry" not in state or len(state["ry"]) != g.cols:
+        state["ry"] = np.random.uniform(-g.rows, g.rows, g.cols).astype(np.float32)
+        state["rsp"] = np.random.uniform(0.3, 2.0, g.cols).astype(np.float32)
+        state["rln"] = np.random.randint(8, 40, g.cols)
+        state["rch"] = np.random.randint(0, len(pal), (g.rows, g.cols))  # pre-assign chars
+
+    speed_mult = speed_base + f.get("bass", 0.3)*speed_beat + f.get("sub_r", 0.3)*3
+    if f.get("beat", 0) > 0: speed_mult *= 2.5
+    state["ry"] += state["rsp"] * speed_mult
+
+    # Reset columns that fall past bottom
+    rst = (state["ry"] - state["rln"]) > g.rows
+    state["ry"][rst] = np.random.uniform(-25, -2, rst.sum())
+
+    # Vectorized draw using fancy indexing
+    ch = np.full((g.rows, g.cols), " ", dtype="U1")
+    co = np.zeros((g.rows, g.cols, 3), dtype=np.uint8)
+    heads = state["ry"].astype(int)
+    for c in range(g.cols):
+        head = heads[c]
+        trail_len = state["rln"][c]
+        for i in range(trail_len):
+            row = head - i
+            if 0 <= row < g.rows:
+                fade = 1.0 - i / trail_len
+                ci = state["rch"][row, c] % len(pal)
+                ch[row, c] = pal[ci]
+                v = fade * bri * 255
+                if i == 0:  # head is bright white-ish
+                    co[row, c] = (int(v*0.9), int(min(255, v*1.1)), int(v*0.9))
+                else:
+                    R, G, B = hsv2rgb_single(hue, 0.7, fade * bri)
+                    co[row, c] = (R, G, B)
+    return ch, co, state
+```
+
+---
+
+## Glitch / Data Effects
+
+### Horizontal Band Displacement
+```python
+def eff_glitch_displace(ch, co, f, intensity=1.0):
+    n_bands = int(8 + f.get("flux", 0.3)*25 + f.get("bdecay", 0)*15) * intensity
+    for _ in range(int(n_bands)):
+        y = random.randint(0, ch.shape[0]-1)
+        h = random.randint(1, int(3 + f.get("sub", 0.3)*8))
+        shift = int((random.random()-0.5) * f.get("rms", 0.3)*40 + f.get("bdecay", 0)*20*(random.random()-0.5))
+        if shift != 0:
+            for row in range(h):
+                rr = y + row
+                if 0 <= rr < ch.shape[0]:
+                    ch[rr] = np.roll(ch[rr], shift)
+                    co[rr] = np.roll(co[rr], shift, axis=0)
+    return ch, co
+```
+
+### Block Corruption
+```python
+def eff_block_corrupt(ch, co, f, char_pool=None, count_base=20):
+    if char_pool is None:
+        char_pool = list(PAL_BLOCKS[4:] + PAL_KATA[2:8])
+    for _ in range(int(count_base + f.get("flux", 0.3)*60 + f.get("bdecay", 0)*40)):
+        bx = random.randint(0, max(1, ch.shape[1]-6))
+        by = random.randint(0, max(1, ch.shape[0]-4))
+        bw, bh = random.randint(2,6), random.randint(1,4)
+        block_char = random.choice(char_pool)
+        # Fill rectangle with single char and random color
+        for r in range(bh):
+            for c in range(bw):
+                rr, cc = by+r, bx+c
+                if 0 <= rr < ch.shape[0] and 0 <= cc < ch.shape[1]:
+                    ch[rr, cc] = block_char
+                    co[rr, cc] = (random.randint(100,255), random.randint(0,100), random.randint(0,80))
+    return ch, co
+```
+
+### Scan Bars (Vertical)
+```python
+def eff_scanbars(ch, co, f, t, n_base=4, chars="|\u2551|!1l"):
+    for bi in range(int(n_base + f.get("himid_r", 0.3)*12)):
+        sx = int((t*50*(1+bi*0.3) + bi*37) % ch.shape[1])
+        for rr in range(ch.shape[0]):
+            if random.random() < 0.7:
+                ch[rr, sx] = random.choice(chars)
+    return ch, co
+```
+
+### Error Messages
+```python
+# Parameterize the error vocabulary per project:
+ERRORS_TECH = ["SEGFAULT","0xDEADBEEF","BUFFER_OVERRUN","PANIC!","NULL_PTR",
+               "CORRUPT","SIGSEGV","ERR_OVERFLOW","STACK_SMASH","BAD_ALLOC"]
+ERRORS_COSMIC = ["VOID_BREACH","ENTROPY_MAX","SINGULARITY","DIMENSION_FAULT",
+                 "REALITY_ERR","TIME_PARADOX","DARK_MATTER_LEAK","QUANTUM_DECOHERE"]
+ERRORS_ORGANIC = ["CELL_DIVISION_ERR","DNA_MISMATCH","MUTATION_OVERFLOW",
+                  "NEURAL_DEADLOCK","SYNAPSE_TIMEOUT","MEMBRANE_BREACH"]
+```
+
+### Hex Data Stream
+```python
+hex_str = "".join(random.choice("0123456789ABCDEF") for _ in range(random.randint(8,20)))
+stamp(ch, co, hex_str, rand_row, rand_col, (0, 160, 80))
+```
+
+---
+
+## Spectrum / Visualization
+
+### Mirrored Spectrum Bars
+```python
+def eff_spectrum(g, f, t, n_bars=64, pal=PAL_BLOCKS, mirror=True):
+    bar_w = max(1, g.cols // n_bars); mid = g.rows // 2
+    band_vals = np.array([f.get("sub",0.3), f.get("bass",0.3), f.get("lomid",0.3),
+                          f.get("mid",0.3), f.get("himid",0.3), f.get("hi",0.3)])
+    ch = np.full((g.rows, g.cols), " ", dtype="U1")
+    co = np.zeros((g.rows, g.cols, 3), dtype=np.uint8)
+    for b in range(n_bars):
+        frac = b / n_bars
+        fi = frac * 5; lo_i = int(fi); hi_i = min(lo_i+1, 5)
+        bval = min(1, (band_vals[lo_i]*(1-fi%1) + band_vals[hi_i]*(fi%1)) * 1.8)
+        height = int(bval * (g.rows//2 - 2))
+        for dy in range(height):
+            hue = (f.get("cent",0.5)*0.3 + frac*0.3 + dy/max(height,1)*0.15) % 1.0
+            ci = pal[min(int(dy/max(height,1)*len(pal)*0.7+len(pal)*0.2), len(pal)-1)]
+            for dc in range(bar_w - (1 if bar_w > 2 else 0)):
+                cc = b*bar_w + dc
+                if 0 <= cc < g.cols:
+                    rows_to_draw = [mid - dy, mid + dy] if mirror else [g.rows - 1 - dy]
+                    for row in rows_to_draw:
+                        if 0 <= row < g.rows:
+                            ch[row, cc] = ci
+                            co[row, cc] = hsv_to_rgb_single(hue, 0.85, 0.5+dy/max(height,1)*0.5)
+    return ch, co
+```
+
+### Waveform
+```python
+def eff_waveform(g, f, t, row_offset=-5, hue=0.1):
+    ch = np.full((g.rows, g.cols), " ", dtype="U1")
+    co = np.zeros((g.rows, g.cols, 3), dtype=np.uint8)
+    for c in range(g.cols):
+        wv = (math.sin(c*0.15+t*5)*f.get("bass",0.3)*0.5
+            + math.sin(c*0.3+t*8)*f.get("mid",0.3)*0.3
+            + math.sin(c*0.6+t*12)*f.get("hi",0.3)*0.15)
+        wr = g.rows + row_offset + int(wv * 4)
+        if 0 <= wr < g.rows:
+            ch[wr, c] = "~"
+            v = int(120 + f.get("rms",0.3)*135)
+            co[wr, c] = [v, int(v*0.7), int(v*0.4)]
+    return ch, co
+```
+
+---
+
+## Fire / Lava
+
+### Fire Columns
+```python
+def eff_fire(g, f, t, n_base=20, hue_base=0.02, hue_range=0.12, pal=PAL_BLOCKS):
+    n_cols = int(n_base + f.get("bass",0.3)*30 + f.get("sub_r",0.3)*20)
+    ch = np.full((g.rows, g.cols), " ", dtype="U1")
+    co = np.zeros((g.rows, g.cols, 3), dtype=np.uint8)
+    for fi in range(n_cols):
+        fx_c = int((fi*g.cols/n_cols + np.sin(t*2+fi*0.7)*3) % g.cols)
+        height = int((f.get("bass",0.3)*0.4 + f.get("sub_r",0.3)*0.3 + f.get("rms",0.3)*0.3) * g.rows * 0.7)
+        for dy in range(min(height, g.rows)):
+            fr = g.rows - 1 - dy
+            frac = dy / max(height, 1)
+            bri = max(0.1, (1 - frac*0.6) * (0.5 + f.get("rms",0.3)*0.5))
+            hue = hue_base + frac * hue_range
+            ci = "\u2588" if frac<0.2 else ("\u2593" if frac<0.4 else ("\u2592" if frac<0.6 else "\u2591"))
+            ch[fr, fx_c] = ci
+            R, G, B = hsv2rgb_single(hue, 0.9, bri)
+            co[fr, fx_c] = (R, G, B)
+    return ch, co
+```
+
+### Ice / Cold Fire (same structure, different hue range)
+```python
+# hue_base=0.55, hue_range=0.15 -- blue to cyan
+# Lower intensity, slower movement
+```
+
+---
+
+## Text Overlays
+
+### Scrolling Ticker
+```python
+def eff_ticker(ch, co, t, text, row, speed=15, color=(80, 100, 140)):
+    off = int(t * speed) % max(len(text), 1)
+    doubled = text + "   " + text
+    stamp(ch, co, doubled[off:off+ch.shape[1]], row, 0, color)
+```
+
+### Beat-Triggered Words
+```python
+def eff_beat_words(ch, co, f, words, row_center=None, color=(255,240,220)):
+    if f.get("beat", 0) > 0:
+        w = random.choice(words)
+        r = (row_center or ch.shape[0]//2) + random.randint(-5,5)
+        stamp(ch, co, w, r, (ch.shape[1]-len(w))//2, color)
+```
+
+### Fading Message Sequence
+```python
+def eff_fading_messages(ch, co, t, elapsed, messages, period=4.0, color_base=(220,220,220)):
+    msg_idx = int(elapsed / period) % len(messages)
+    phase = elapsed % period
+    fade = max(0, min(1.0, phase) * min(1.0, period - phase))
+    if fade > 0.05:
+        v = fade
+        msg = messages[msg_idx]
+        cr, cg, cb = [int(c * v) for c in color_base]
+        stamp(ch, co, msg, ch.shape[0]//2, (ch.shape[1]-len(msg))//2, (cr, cg, cb))
+```
+
+---
+
+## Screen Shake
+Shift entire char/color arrays on beat:
+```python
+def eff_shake(ch, co, f, x_amp=6, y_amp=3):
+    shake_x = int(f.get("sub",0.3)*x_amp*(random.random()-0.5)*2 + f.get("bdecay",0)*4*(random.random()-0.5)*2)
+    shake_y = int(f.get("bass",0.3)*y_amp*(random.random()-0.5)*2)
+    if abs(shake_x) > 0:
+        ch = np.roll(ch, shake_x, axis=1)
+        co = np.roll(co, shake_x, axis=1)
+    if abs(shake_y) > 0:
+        ch = np.roll(ch, shake_y, axis=0)
+        co = np.roll(co, shake_y, axis=0)
+    return ch, co
+```
+
+---
+
+## Composable Effect System
+
+The real creative power comes from **composition**. There are three levels:
+
+### Level 1: Character-Level Layering
+
+Stack multiple effects as `(chars, colors)` layers:
+
+```python
+class LayerStack(EffectNode):
+    """Render effects bottom-to-top with character-level compositing."""
+    def add(self, effect, alpha=1.0):
+        """alpha < 1.0 = probabilistic override (sparse overlay)."""
+        self.layers.append((effect, alpha))
+
+# Usage:
+stack = LayerStack()
+stack.add(bg_effect)           # base — fills screen
+stack.add(main_effect)         # overlay on top (space chars = transparent)
+stack.add(particle_effect)     # sparse overlay on top of that
+ch, co = stack.render(g, f, t, S)
+```
+
+### Level 2: Pixel-Level Blending
+
+After rendering to canvases, blend with Photoshop-style modes:
+
+```python
+class PixelBlendStack:
+    """Stack canvases with blend modes for complex compositing."""
+    def add(self, canvas, mode="normal", opacity=1.0)
+    def composite(self) -> canvas
+
+# Usage:
+pbs = PixelBlendStack()
+pbs.add(canvas_a)                        # base
+pbs.add(canvas_b, "screen", 0.7)        # additive glow
+pbs.add(canvas_c, "difference", 0.5)    # psychedelic interference
+result = pbs.composite()
+```
+
+### Level 3: Temporal Feedback
+
+Feed previous frame back into current frame for recursive effects:
+
+```python
+fb = FeedbackBuffer()
+for each frame:
+    canvas = render_current()
+    canvas = fb.apply(canvas, decay=0.8, blend="screen",
+                      transform="zoom", transform_amt=0.015, hue_shift=0.02)
+```
+
+### Effect Nodes — Uniform Interface
+
+In the v2 protocol, effect nodes are used **inside** scene functions. The scene function itself returns a canvas. Effect nodes produce intermediate `(chars, colors)` that are rendered to canvas via the grid's `.render()` method or `_render_vf()`.
+
+```python
+class EffectNode:
+    def render(self, g, f, t, S) -> (chars, colors)
+
+# Concrete implementations:
+class ValueFieldEffect(EffectNode):
+    """Wraps a value field function + hue field function + palette."""
+    def __init__(self, val_fn, hue_fn, pal=PAL_DEFAULT, sat=0.7)
+
+class LambdaEffect(EffectNode):
+    """Wrap any (g,f,t,S) -> (ch,co) function."""
+    def __init__(self, fn)
+
+class ConditionalEffect(EffectNode):
+    """Switch effects based on audio features."""
+    def __init__(self, condition, if_true, if_false=None)
+```
+
+### Value Field Generators (Atomic Building Blocks)
+
+These produce float32 arrays `(rows, cols)` in range [0,1]. They are the raw visual patterns. All have signature `(g, f, t, S, **params) -> float32 array`.
+
+```python
+def vf_sinefield(g, f, t, S, bri=0.5,
+                 freq=(0.13, 0.17, 0.07, 0.09), speed=(0.5, -0.4, -0.3, 0.2)):
+    """Layered sine field. General purpose background/texture."""
+    v1 = np.sin(g.cc*freq[0] + t*speed[0]) * np.sin(g.rr*freq[1] - t*speed[1]) * 0.5 + 0.5
+    v2 = np.sin(g.cc*freq[2] - t*speed[2] + g.rr*freq[3]) * 0.4 + 0.5
+    v3 = np.sin(g.dist_n*5 + t*0.2) * 0.3 + 0.4
+    return np.clip((v1*0.35 + v2*0.35 + v3*0.3) * bri * (0.6 + f.get("rms",0.3)*0.6), 0, 1)
+
+def vf_smooth_noise(g, f, t, S, octaves=3, bri=0.5):
+    """Multi-octave sine approximation of Perlin noise."""
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for i in range(octaves):
+        freq = 0.05 * (2 ** i); amp = 0.5 / (i + 1)
+        phase = t * (0.3 + i * 0.2)
+        val = val + np.sin(g.cc*freq + phase) * np.cos(g.rr*freq*0.7 - phase*0.5) * amp
+    return np.clip(val * 0.5 + 0.5, 0, 1) * bri
+
+def vf_rings(g, f, t, S, n_base=6, spacing_base=4):
+    """Concentric rings, bass-driven count and wobble."""
+    n = int(n_base + f.get("sub_r",0.3)*25 + f.get("bass",0.3)*10)
+    sp = spacing_base + f.get("bass_r",0.3)*7 + f.get("rms",0.3)*3
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for ri in range(n):
+        rad = (ri+1)*sp + f.get("bdecay",0)*15
+        wobble = f.get("mid_r",0.3)*5*np.sin(g.angle*3+t*4)
+        rd = np.abs(g.dist - rad - wobble)
+        th = 1 + f.get("sub",0.3)*3
+        val = np.maximum(val, np.clip((1 - rd/th) * (0.4 + f.get("bass",0.3)*0.8), 0, 1))
+    return val
+
+def vf_spiral(g, f, t, S, n_arms=3, tightness=2.5):
+    """Logarithmic spiral arms."""
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for ai in range(n_arms):
+        offset = ai * 2*np.pi / n_arms
+        log_r = np.log(g.dist + 1) * tightness
+        arm_phase = g.angle + offset - log_r + t * 0.8
+        arm_val = np.clip(np.cos(arm_phase * n_arms) * 0.6 + 0.2, 0, 1)
+        arm_val *= (0.4 + f.get("rms",0.3)*0.6) * np.clip(1 - g.dist_n*0.5, 0.2, 1)
+        val = np.maximum(val, arm_val)
+    return val
+
+def vf_tunnel(g, f, t, S, speed=3.0, complexity=6):
+    """Tunnel depth effect — infinite zoom feeling."""
+    tunnel_d = 1.0 / (g.dist_n + 0.1)
+    v1 = np.sin(tunnel_d*2 - t*speed) * 0.45 + 0.55
+    v2 = np.sin(g.angle*complexity + tunnel_d*1.5 - t*2) * 0.35 + 0.55
+    return np.clip(v1*0.5 + v2*0.5, 0, 1)
+
+def vf_vortex(g, f, t, S, twist=3.0):
+    """Twisting radial pattern — distance modulates angle."""
+    twisted = g.angle + g.dist_n * twist * np.sin(t * 0.5)
+    val = np.sin(twisted * 4 - t * 2) * 0.5 + 0.5
+    return np.clip(val * (0.5 + f.get("bass",0.3)*0.8), 0, 1)
+
+def vf_interference(g, f, t, S, n_waves=6):
+    """Overlapping sine waves creating moire patterns."""
+    drivers = ["mid_r", "himid_r", "bass_r", "lomid_r", "hi_r", "sub_r"]
+    vals = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for i in range(min(n_waves, len(drivers))):
+        angle = i * np.pi / n_waves
+        freq = 0.06 + i * 0.03; sp = 0.5 + i * 0.3
+        proj = g.cc * np.cos(angle) + g.rr * np.sin(angle)
+        vals = vals + np.sin(proj*freq + t*sp) * f.get(drivers[i], 0.3) * 2.5
+    return np.clip(vals * 0.12 + 0.45, 0.1, 1)
+
+def vf_aurora(g, f, t, S, n_bands=3):
+    """Horizontal aurora bands."""
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for i in range(n_bands):
+        fr = 0.08 + i*0.04; fc = 0.012 + i*0.008
+        sr = 0.7 + i*0.3; sc = 0.18 + i*0.12
+        val = val + np.sin(g.rr*fr + t*sr) * np.sin(g.cc*fc + t*sc) * (0.6/n_bands)
+    return np.clip(val * (f.get("lomid_r",0.3)*3 + 0.2), 0, 0.7)
+
+def vf_ripple(g, f, t, S, sources=None, freq=0.3, damping=0.02):
+    """Concentric ripples from point sources."""
+    if sources is None: sources = [(0.5, 0.5)]
+    val = np.zeros((g.rows, g.cols), dtype=np.float32)
+    for ry, rx in sources:
+        dy = g.rr - g.rows*ry; dx = g.cc - g.cols*rx
+        d = np.sqrt(dy**2 + dx**2)
+        val = val + np.sin(d*freq - t*4) * np.exp(-d*damping) * 0.5
+    return np.clip(val + 0.5, 0, 1)
+
+def vf_plasma(g, f, t, S):
+    """Classic plasma: sum of sines at different orientations and speeds."""
+    v = np.sin(g.cc * 0.03 + t * 0.7) * 0.5
+    v = v + np.sin(g.rr * 0.04 - t * 0.5) * 0.4
+    v = v + np.sin((g.cc * 0.02 + g.rr * 0.03) + t * 0.3) * 0.3
+    v = v + np.sin(g.dist_n * 4 - t * 0.8) * 0.3
+    return np.clip(v * 0.5 + 0.5, 0, 1)
+
+def vf_diamond(g, f, t, S, freq=0.15):
+    """Diamond/checkerboard pattern."""
+    val = np.abs(np.sin(g.cc * freq + t * 0.5)) * np.abs(np.sin(g.rr * freq * 1.2 - t * 0.3))
+    return np.clip(val * (0.6 + f.get("rms",0.3)*0.8), 0, 1)
+
+def vf_noise_static(g, f, t, S, density=0.4):
+    """Random noise — different each frame. Non-deterministic."""
+    return np.random.random((g.rows, g.cols)).astype(np.float32) * density * (0.5 + f.get("rms",0.3)*0.5)
+```
+
+### Hue Field Generators (Color Mapping)
+
+These produce float32 hue arrays [0,1]. Independently combinable with any value field. Each is a factory returning a closure with signature `(g, f, t, S) -> float32 array`. Can also be a plain float for fixed hue.
+
+```python
+def hf_fixed(hue):
+    """Single hue everywhere."""
+    def fn(g, f, t, S):
+        return np.full((g.rows, g.cols), hue, dtype=np.float32)
+    return fn
+
+def hf_angle(offset=0.0):
+    """Hue mapped to angle from center — rainbow wheel."""
+    def fn(g, f, t, S):
+        return (g.angle / (2 * np.pi) + offset + t * 0.05) % 1.0
+    return fn
+
+def hf_distance(base=0.5, scale=0.02):
+    """Hue mapped to distance from center."""
+    def fn(g, f, t, S):
+        return (base + g.dist * scale + t * 0.03) % 1.0
+    return fn
+
+def hf_time_cycle(speed=0.1):
+    """Hue cycles uniformly over time."""
+    def fn(g, f, t, S):
+        return np.full((g.rows, g.cols), (t * speed) % 1.0, dtype=np.float32)
+    return fn
+
+def hf_audio_cent():
+    """Hue follows spectral centroid — timbral color shifting."""
+    def fn(g, f, t, S):
+        return np.full((g.rows, g.cols), f.get("cent", 0.5) * 0.3, dtype=np.float32)
+    return fn
+
+def hf_gradient_h(start=0.0, end=1.0):
+    """Left-to-right hue gradient."""
+    def fn(g, f, t, S):
+        h = np.broadcast_to(
+            start + (g.cc / g.cols) * (end - start),
+            (g.rows, g.cols)
+        ).copy()  # .copy() is CRITICAL — see troubleshooting.md
+        return h % 1.0
+    return fn
+
+def hf_gradient_v(start=0.0, end=1.0):
+    """Top-to-bottom hue gradient."""
+    def fn(g, f, t, S):
+        h = np.broadcast_to(
+            start + (g.rr / g.rows) * (end - start),
+            (g.rows, g.cols)
+        ).copy()
+        return h % 1.0
+    return fn
+
+def hf_plasma(speed=0.3):
+    """Plasma-style hue field — organic color variation."""
+    def fn(g, f, t, S):
+        return (np.sin(g.cc*0.02 + t*speed)*0.5 + np.sin(g.rr*0.015 + t*speed*0.7)*0.5) % 1.0
+    return fn
+```
+
+### Combining Value Fields
+
+The combinatorial explosion comes from mixing value fields with math:
+
+```python
+# Multiplication = intersection (only shows where both have brightness)
+combined = vf_plasma(g,f,t,S) * vf_vortex(g,f,t,S)
+
+# Addition = union (shows both, clips at 1.0)
+combined = np.clip(vf_rings(g,f,t,S) + vf_spiral(g,f,t,S), 0, 1)
+
+# Interference = beat pattern (shows XOR-like patterns)
+combined = np.abs(vf_plasma(g,f,t,S) - vf_tunnel(g,f,t,S))
+
+# Modulation = one effect shapes the other
+combined = vf_rings(g,f,t,S) * (0.3 + 0.7 * vf_plasma(g,f,t,S))
+
+# Maximum = shows the brightest of two effects
+combined = np.maximum(vf_spiral(g,f,t,S), vf_aurora(g,f,t,S))
+```
+
+### Full Scene Example (v2 — Canvas Return)
+
+A v2 scene function composes effects internally and returns a pixel canvas:
+
+```python
+def scene_complex(r, f, t, S):
+    """v2 scene function: returns canvas (uint8 H,W,3).
+    r = Renderer, f = audio features, t = time, S = persistent state dict."""
+    g = r.grids["md"]
+    rows, cols = g.rows, g.cols
+    
+    # 1. Value field composition
+    plasma = vf_plasma(g, f, t, S)
+    vortex = vf_vortex(g, f, t, S, twist=4.0)
+    combined = np.clip(plasma * 0.6 + vortex * 0.5 + plasma * vortex * 0.4, 0, 1)
+    
+    # 2. Color from hue field
+    h = (hf_angle(0.3)(g,f,t,S) * 0.5 + hf_time_cycle(0.08)(g,f,t,S) * 0.5) % 1.0
+    
+    # 3. Render to canvas via _render_vf helper
+    canvas = _render_vf(g, combined, h, sat=0.75, pal=PAL_DENSE)
+    
+    # 4. Optional: blend a second layer
+    overlay = _render_vf(r.grids["sm"], vf_rings(r.grids["sm"],f,t,S),
+                         hf_fixed(0.6)(r.grids["sm"],f,t,S), pal=PAL_BLOCK)
+    canvas = blend_canvas(canvas, overlay, "screen", 0.4)
+    
+    return canvas
+    
+# In the render_clip() loop (handled by the framework):
+# canvas = scene_fn(r, f, t, S)
+# canvas = tonemap(canvas, gamma=scene_gamma)
+# canvas = feedback.apply(canvas, ...)
+# canvas = shader_chain.apply(canvas, f=f, t=t)
+# pipe.stdin.write(canvas.tobytes())
+```
+
+Vary the **value field combo**, **hue field**, **palette**, **blend modes**, **feedback config**, and **shader chain** per section for maximum visual variety. With 12 value fields × 8 hue fields × 14 palettes × 20 blend modes × 7 feedback transforms × 38 shaders, the combinations are effectively infinite.
--- a/skills/creative/ascii-video/references/inputs.md
+++ b/skills/creative/ascii-video/references/inputs.md
@@ -0,0 +1,407 @@
+# Input Sources
+
+## Audio Analysis
+
+### Loading
+
+```python
+tmp = tempfile.mktemp(suffix=".wav")
+subprocess.run(["ffmpeg", "-y", "-i", input_path, "-ac", "1", "-ar", "22050",
+                "-sample_fmt", "s16", tmp], capture_output=True, check=True)
+with wave.open(tmp) as wf:
+    sr = wf.getframerate()
+    raw = wf.readframes(wf.getnframes())
+samples = np.frombuffer(raw, dtype=np.int16).astype(np.float32) / 32768.0
+```
+
+### Per-Frame FFT
+
+```python
+hop = sr // fps          # samples per frame
+win = hop * 2            # analysis window (2x hop for overlap)
+window = np.hanning(win)
+freqs = rfftfreq(win, 1.0 / sr)
+
+bands = {
+    "sub":   (freqs >= 20)  & (freqs < 80),
+    "bass":  (freqs >= 80)  & (freqs < 250),
+    "lomid": (freqs >= 250) & (freqs < 500),
+    "mid":   (freqs >= 500) & (freqs < 2000),
+    "himid": (freqs >= 2000)& (freqs < 6000),
+    "hi":    (freqs >= 6000),
+}
+```
+
+For each frame: extract chunk, apply window, FFT, compute band energies.
+
+### Feature Set
+
+| Feature | Formula | Controls |
+|---------|---------|----------|
+| `rms` | `sqrt(mean(chunk²))` | Overall loudness/energy |
+| `sub`..`hi` | `sqrt(mean(band_magnitudes²))` | Per-band energy |
+| `centroid` | `sum(freq*mag) / sum(mag)` | Brightness/timbre |
+| `flatness` | `geomean(mag) / mean(mag)` | Noise vs tone |
+| `flux` | `sum(max(0, mag - prev_mag))` | Transient strength |
+| `sub_r`..`hi_r` | `band / sum(all_bands)` | Spectral shape (volume-independent) |
+| `cent_d` | `abs(gradient(centroid))` | Timbral change rate |
+| `beat` | Flux peak detection | Binary beat onset |
+| `bdecay` | Exponential decay from beats | Smooth beat pulse (0→1→0) |
+
+**Band ratios are critical** — they decouple spectral shape from volume, so a quiet bass section and a loud bass section both read as "bassy" rather than just "loud" vs "quiet".
+
+### Smoothing
+
+EMA prevents visual jitter:
+
+```python
+def ema(arr, alpha):
+    out = np.empty_like(arr); out[0] = arr[0]
+    for i in range(1, len(arr)):
+        out[i] = alpha * arr[i] + (1 - alpha) * out[i-1]
+    return out
+
+# Slow-moving features (alpha=0.12): centroid, flatness, band ratios, cent_d
+# Fast-moving features (alpha=0.3): rms, flux, raw bands
+```
+
+### Beat Detection
+
+```python
+flux_smooth = np.convolve(flux, np.ones(5)/5, mode="same")
+peaks, _ = signal.find_peaks(flux_smooth, height=0.15, distance=fps//5, prominence=0.05)
+
+beat = np.zeros(n_frames)
+bdecay = np.zeros(n_frames, dtype=np.float32)
+for p in peaks:
+    beat[p] = 1.0
+    for d in range(fps // 2):
+        if p + d < n_frames:
+            bdecay[p + d] = max(bdecay[p + d], math.exp(-d * 2.5 / (fps // 2)))
+```
+
+`bdecay` gives smooth 0→1→0 pulse per beat, decaying over ~0.5s. Use for flash/glitch/mirror triggers.
+
+### Normalization
+
+After computing all frames, normalize each feature to 0-1:
+
+```python
+for k in features:
+    a = features[k]
+    lo, hi = a.min(), a.max()
+    features[k] = (a - lo) / (hi - lo + 1e-10)
+```
+
+## Video Sampling
+
+### Frame Extraction
+
+```python
+# Method 1: ffmpeg pipe (memory efficient)
+cmd = ["ffmpeg", "-i", input_video, "-f", "rawvideo", "-pix_fmt", "rgb24",
+       "-s", f"{target_w}x{target_h}", "-r", str(fps), "-"]
+pipe = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
+frame_size = target_w * target_h * 3
+for fi in range(n_frames):
+    raw = pipe.stdout.read(frame_size)
+    if len(raw) < frame_size: break
+    frame = np.frombuffer(raw, dtype=np.uint8).reshape(target_h, target_w, 3)
+    # process frame...
+
+# Method 2: OpenCV (if available)
+cap = cv2.VideoCapture(input_video)
+```
+
+### Luminance-to-Character Mapping
+
+Convert video pixels to ASCII characters based on brightness:
+
+```python
+def frame_to_ascii(frame_rgb, grid, pal=PAL_DEFAULT):
+    """Convert video frame to character + color arrays."""
+    rows, cols = grid.rows, grid.cols
+    # Resize frame to grid dimensions
+    small = np.array(Image.fromarray(frame_rgb).resize((cols, rows), Image.LANCZOS))
+    # Luminance
+    lum = (0.299 * small[:,:,0] + 0.587 * small[:,:,1] + 0.114 * small[:,:,2]) / 255.0
+    # Map to chars
+    chars = val2char(lum, lum > 0.02, pal)
+    # Colors: use source pixel colors, scaled by luminance for visibility
+    colors = np.clip(small * np.clip(lum[:,:,None] * 1.5 + 0.3, 0.3, 1), 0, 255).astype(np.uint8)
+    return chars, colors
+```
+
+### Edge-Weighted Character Mapping
+
+Use edge detection for more detail in contour regions:
+
+```python
+def frame_to_ascii_edges(frame_rgb, grid, pal=PAL_DEFAULT, edge_pal=PAL_BOX):
+    gray = np.mean(frame_rgb, axis=2)
+    small_gray = resize(gray, (grid.rows, grid.cols))
+    lum = small_gray / 255.0
+
+    # Sobel edge detection
+    gx = np.abs(small_gray[:, 2:] - small_gray[:, :-2])
+    gy = np.abs(small_gray[2:, :] - small_gray[:-2, :])
+    edge = np.zeros_like(small_gray)
+    edge[:, 1:-1] += gx; edge[1:-1, :] += gy
+    edge = np.clip(edge / edge.max(), 0, 1)
+
+    # Edge regions get box drawing chars, flat regions get brightness chars
+    is_edge = edge > 0.15
+    chars = val2char(lum, lum > 0.02, pal)
+    edge_chars = val2char(edge, is_edge, edge_pal)
+    chars[is_edge] = edge_chars[is_edge]
+
+    return chars, colors
+```
+
+### Motion Detection
+
+Detect pixel changes between frames for motion-reactive effects:
+
+```python
+prev_frame = None
+def compute_motion(frame):
+    global prev_frame
+    if prev_frame is None:
+        prev_frame = frame.astype(np.float32)
+        return np.zeros(frame.shape[:2])
+    diff = np.abs(frame.astype(np.float32) - prev_frame).mean(axis=2)
+    prev_frame = frame.astype(np.float32) * 0.7 + prev_frame * 0.3  # smoothed
+    return np.clip(diff / 30.0, 0, 1)  # normalized motion map
+```
+
+Use motion map to drive particle emission, glitch intensity, or character density.
+
+### Video Feature Extraction
+
+Per-frame features analogous to audio features, for driving effects:
+
+```python
+def analyze_video_frame(frame_rgb):
+    gray = np.mean(frame_rgb, axis=2)
+    return {
+        "brightness": gray.mean() / 255.0,
+        "contrast": gray.std() / 128.0,
+        "edge_density": compute_edge_density(gray),
+        "motion": compute_motion(frame_rgb).mean(),
+        "dominant_hue": compute_dominant_hue(frame_rgb),
+        "color_variance": compute_color_variance(frame_rgb),
+    }
+```
+
+## Image Sequence
+
+### Static Image to ASCII
+
+Same as single video frame conversion. For animated sequences:
+
+```python
+import glob
+frames = sorted(glob.glob("frames/*.png"))
+for fi, path in enumerate(frames):
+    img = np.array(Image.open(path).resize((VW, VH)))
+    chars, colors = frame_to_ascii(img, grid, pal)
+```
+
+### Image as Texture Source
+
+Use an image as a background texture that effects modulate:
+
+```python
+def load_texture(path, grid):
+    img = np.array(Image.open(path).resize((grid.cols, grid.rows)))
+    lum = np.mean(img, axis=2) / 255.0
+    return lum, img  # luminance for char mapping, RGB for colors
+```
+
+## Text / Lyrics
+
+### SRT Parsing
+
+```python
+import re
+def parse_srt(path):
+    """Returns [(start_sec, end_sec, text), ...]"""
+    entries = []
+    with open(path) as f:
+        content = f.read()
+    blocks = content.strip().split("\n\n")
+    for block in blocks:
+        lines = block.strip().split("\n")
+        if len(lines) >= 3:
+            times = lines[1]
+            m = re.match(r"(\d+):(\d+):(\d+),(\d+) --> (\d+):(\d+):(\d+),(\d+)", times)
+            if m:
+                g = [int(x) for x in m.groups()]
+                start = g[0]*3600 + g[1]*60 + g[2] + g[3]/1000
+                end = g[4]*3600 + g[5]*60 + g[6] + g[7]/1000
+                text = " ".join(lines[2:])
+                entries.append((start, end, text))
+    return entries
+```
+
+### Lyrics Display Modes
+
+- **Typewriter**: characters appear left-to-right over the time window
+- **Fade-in**: whole line fades from dark to bright
+- **Flash**: appear instantly on beat, fade out
+- **Scatter**: characters start at random positions, converge to final position
+- **Wave**: text follows a sine wave path
+
+```python
+def lyrics_typewriter(ch, co, text, row, col, t, t_start, t_end, color):
+    """Reveal characters progressively over time window."""
+    progress = np.clip((t - t_start) / (t_end - t_start), 0, 1)
+    n_visible = int(len(text) * progress)
+    stamp(ch, co, text[:n_visible], row, col, color)
+```
+
+## Generative (No Input)
+
+For pure generative ASCII art, the "features" dict is synthesized from time:
+
+```python
+def synthetic_features(t, bpm=120):
+    """Generate audio-like features from time alone."""
+    beat_period = 60.0 / bpm
+    beat_phase = (t % beat_period) / beat_period
+    return {
+        "rms": 0.5 + 0.3 * math.sin(t * 0.5),
+        "bass": 0.5 + 0.4 * math.sin(t * 2 * math.pi / beat_period),
+        "sub": 0.3 + 0.3 * math.sin(t * 0.8),
+        "mid": 0.4 + 0.3 * math.sin(t * 1.3),
+        "hi": 0.3 + 0.2 * math.sin(t * 2.1),
+        "cent": 0.5 + 0.2 * math.sin(t * 0.3),
+        "flat": 0.4,
+        "flux": 0.3 + 0.2 * math.sin(t * 3),
+        "beat": 1.0 if beat_phase < 0.05 else 0.0,
+        "bdecay": max(0, 1.0 - beat_phase * 4),
+        # ratios
+        "sub_r": 0.2, "bass_r": 0.25, "lomid_r": 0.15,
+        "mid_r": 0.2, "himid_r": 0.12, "hi_r": 0.08,
+        "cent_d": 0.1,
+    }
+```
+
+## TTS Integration
+
+For narrated videos (testimonials, quotes, storytelling), generate speech audio per segment and mix with background music.
+
+### ElevenLabs Voice Generation
+
+```python
+import requests
+
+def generate_tts(text, voice_id, api_key, output_path, model="eleven_multilingual_v2"):
+    """Generate TTS audio via ElevenLabs API."""
+    url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}"
+    headers = {"xi-api-key": api_key, "Content-Type": "application/json"}
+    data = {"text": text, "model_id": model,
+            "voice_settings": {"stability": 0.5, "similarity_boost": 0.75}}
+    resp = requests.post(url, json=data, headers=headers, timeout=30)
+    resp.raise_for_status()
+    with open(output_path, "wb") as f:
+        f.write(resp.content)
+```
+
+### Voice Assignment
+
+Use multiple voices for variety. Shuffle deterministically so re-runs are consistent:
+
+```python
+import random as _rng
+
+def assign_voices(n_quotes, voice_pool, seed=42):
+    """Assign a different voice to each quote, cycling if needed."""
+    r = _rng.Random(seed)
+    shuffled = list(voice_pool)
+    r.shuffle(shuffled)
+    return [shuffled[i % len(shuffled)] for i in range(n_quotes)]
+```
+
+### Pronunciation Control
+
+TTS text should be separate from display text. Common fixes:
+- Brand names: spell phonetically ("Nous" -> "Noose", "nginx" -> "engine-x")
+- Abbreviations: expand ("API" -> "A P I", "CLI" -> "C L I")
+- Technical terms: add phonetic hints
+
+```python
+QUOTES = [("Display text here", "Author")]
+QUOTES_TTS = ["TTS text with phonetic spelling here"]
+# Keep both arrays in sync -- same indices
+```
+
+### Audio Pipeline
+
+1. Generate individual TTS clips (MP3/WAV per quote)
+2. Get duration of each clip
+3. Calculate timing: speech start/end per quote with gaps
+4. Concatenate into single TTS track with silence padding
+5. Mix with background music
+
+```python
+def build_tts_track(tts_clips, target_duration, gap_seconds=2.0):
+    """Concatenate TTS clips with gaps, pad to target duration."""
+    # Get durations
+    durations = []
+    for clip in tts_clips:
+        result = subprocess.run(
+            ["ffprobe", "-v", "error", "-show_entries", "format=duration",
+             "-of", "csv=p=0", clip],
+            capture_output=True, text=True)
+        durations.append(float(result.stdout.strip()))
+    
+    # Calculate timing
+    total_speech = sum(durations)
+    total_gaps = target_duration - total_speech
+    gap = max(0.5, total_gaps / (len(tts_clips) + 1))
+    
+    timing = []  # (start, end, quote_index)
+    t = gap  # start after initial gap
+    for i, dur in enumerate(durations):
+        timing.append((t, t + dur, i))
+        t += dur + gap
+    
+    # Concatenate with ffmpeg
+    # ... silence padding + concat filter
+    return timing
+```
+
+### Audio Mixing
+
+Mix TTS (center) with background music (wide stereo, low volume):
+
+```python
+def mix_audio(tts_path, bgm_path, output_path, bgm_volume=0.15):
+    """Mix TTS centered with BGM panned wide stereo."""
+    cmd = [
+        "ffmpeg", "-y",
+        "-i", tts_path,   # mono TTS
+        "-i", bgm_path,   # stereo BGM
+        "-filter_complex",
+        f"[0:a]aformat=sample_fmts=fltp:sample_rates=44100:channel_layouts=mono,"
+        f"pan=stereo|c0=c0|c1=c0[tts];"  # TTS center
+        f"[1:a]loudnorm=I=-16:TP=-1.5:LRA=11,"
+        f"volume={bgm_volume},"
+        f"extrastereo=2.5[bgm];"  # BGM wide stereo
+        f"[tts][bgm]amix=inputs=2:duration=longest[out]",
+        "-map", "[out]", "-c:a", "pcm_s16le", output_path
+    ]
+    subprocess.run(cmd, capture_output=True, check=True)
+```
+
+### Feature Analysis on Mixed Audio
+
+Run the standard audio analysis (FFT, beat detection) on the final mixed track so visual effects react to both TTS and music:
+
+```python
+# Analyze mixed_final.wav (not individual tracks)
+features = analyze_audio("mixed_final.wav", fps=24)
+```
+
+This means visuals will pulse with both the music beats and the speech energy -- creating natural synchronization.
--- a/skills/creative/ascii-video/references/optimization.md
+++ b/skills/creative/ascii-video/references/optimization.md
@@ -0,0 +1,435 @@
+# Optimization Reference
+
+## Hardware Detection
+
+Detect the user's hardware at script startup and adapt rendering parameters automatically. Never hardcode worker counts or resolution.
+
+### CPU and Memory Detection
+
+```python
+import multiprocessing
+import platform
+import shutil
+import os
+
+def detect_hardware():
+    """Detect hardware capabilities and return render config."""
+    cpu_count = multiprocessing.cpu_count()
+    
+    # Leave 1-2 cores free for OS + ffmpeg encoding
+    if cpu_count >= 16:
+        workers = cpu_count - 2
+    elif cpu_count >= 8:
+        workers = cpu_count - 1
+    elif cpu_count >= 4:
+        workers = cpu_count - 1
+    else:
+        workers = max(1, cpu_count)
+    
+    # Memory detection (platform-specific)
+    try:
+        if platform.system() == "Darwin":
+            import subprocess
+            mem_bytes = int(subprocess.check_output(["sysctl", "-n", "hw.memsize"]).strip())
+        elif platform.system() == "Linux":
+            with open("/proc/meminfo") as f:
+                for line in f:
+                    if line.startswith("MemTotal"):
+                        mem_bytes = int(line.split()[1]) * 1024
+                        break
+        else:
+            mem_bytes = 8 * 1024**3  # assume 8GB on unknown
+    except Exception:
+        mem_bytes = 8 * 1024**3
+
+    mem_gb = mem_bytes / (1024**3)
+    
+    # Each worker uses ~50-150MB depending on grid sizes
+    # Cap workers if memory is tight
+    mem_per_worker_mb = 150
+    max_workers_by_mem = int(mem_gb * 1024 * 0.6 / mem_per_worker_mb)  # use 60% of RAM
+    workers = min(workers, max_workers_by_mem)
+    
+    # ffmpeg availability and codec support
+    has_ffmpeg = shutil.which("ffmpeg") is not None
+    
+    return {
+        "cpu_count": cpu_count,
+        "workers": workers,
+        "mem_gb": mem_gb,
+        "platform": platform.system(),
+        "arch": platform.machine(),
+        "has_ffmpeg": has_ffmpeg,
+    }
+```
+
+### Adaptive Quality Profiles
+
+Scale resolution, FPS, CRF, and grid density based on hardware:
+
+```python
+def quality_profile(hw, target_duration_s, user_preference="auto"):
+    """
+    Returns render settings adapted to hardware.
+    user_preference: "auto", "draft", "preview", "production", "max"
+    """
+    if user_preference == "draft":
+        return {"vw": 960, "vh": 540, "fps": 12, "crf": 28, "workers": min(4, hw["workers"]),
+                "grid_scale": 0.5, "shaders": "minimal", "particles_max": 200}
+    
+    if user_preference == "preview":
+        return {"vw": 1280, "vh": 720, "fps": 15, "crf": 25, "workers": hw["workers"],
+                "grid_scale": 0.75, "shaders": "standard", "particles_max": 500}
+    
+    if user_preference == "max":
+        return {"vw": 3840, "vh": 2160, "fps": 30, "crf": 15, "workers": hw["workers"],
+                "grid_scale": 2.0, "shaders": "full", "particles_max": 3000}
+    
+    # "production" or "auto"
+    # Auto-detect: estimate render time, downgrade if it would take too long
+    n_frames = int(target_duration_s * 24)
+    est_seconds_per_frame = 0.18  # ~180ms at 1080p
+    est_total_s = n_frames * est_seconds_per_frame / max(1, hw["workers"])
+    
+    if hw["mem_gb"] < 4 or hw["cpu_count"] <= 2:
+        # Low-end: 720p, 15fps
+        return {"vw": 1280, "vh": 720, "fps": 15, "crf": 23, "workers": hw["workers"],
+                "grid_scale": 0.75, "shaders": "standard", "particles_max": 500}
+    
+    if est_total_s > 3600:  # would take over an hour
+        # Downgrade to 720p to speed up
+        return {"vw": 1280, "vh": 720, "fps": 24, "crf": 20, "workers": hw["workers"],
+                "grid_scale": 0.75, "shaders": "standard", "particles_max": 800}
+    
+    # Standard production: 1080p 24fps
+    return {"vw": 1920, "vh": 1080, "fps": 24, "crf": 20, "workers": hw["workers"],
+            "grid_scale": 1.0, "shaders": "full", "particles_max": 1200}
+
+
+def apply_quality_profile(profile):
+    """Set globals from quality profile."""
+    global VW, VH, FPS, N_WORKERS
+    VW = profile["vw"]
+    VH = profile["vh"]
+    FPS = profile["fps"]
+    N_WORKERS = profile["workers"]
+    # Grid sizes scale with resolution
+    # CRF passed to ffmpeg encoder
+    # Shader set determines which post-processing is active
+```
+
+### CLI Integration
+
+```python
+parser = argparse.ArgumentParser()
+parser.add_argument("--quality", choices=["draft", "preview", "production", "max", "auto"],
+                    default="auto", help="Render quality preset")
+parser.add_argument("--workers", type=int, default=0, help="Override worker count (0=auto)")
+parser.add_argument("--resolution", type=str, default="", help="Override resolution e.g. 1280x720")
+args = parser.parse_args()
+
+hw = detect_hardware()
+if args.workers > 0:
+    hw["workers"] = args.workers
+profile = quality_profile(hw, target_duration, args.quality)
+if args.resolution:
+    w, h = args.resolution.split("x")
+    profile["vw"], profile["vh"] = int(w), int(h)
+apply_quality_profile(profile)
+
+log(f"Hardware: {hw['cpu_count']} cores, {hw['mem_gb']:.1f}GB RAM, {hw['platform']}")
+log(f"Render:   {profile['vw']}x{profile['vh']} @{profile['fps']}fps, "
+    f"CRF {profile['crf']}, {profile['workers']} workers")
+```
+
+## Performance Budget
+
+Target: 100-200ms per frame (5-10 fps single-threaded, 40-80 fps across 8 workers).
+
+| Component | Time | Notes |
+|-----------|------|-------|
+| Feature extraction | 1-5ms | Pre-computed for all frames before render |
+| Effect function | 2-15ms | Vectorized numpy, avoid Python loops |
+| Character render | 80-150ms | **Bottleneck** -- per-cell Python loop |
+| Shader pipeline | 5-25ms | Depends on active shaders |
+| ffmpeg encode | ~5ms | Amortized by pipe buffering |
+
+## Bitmap Pre-Rasterization
+
+Rasterize every character at init, not per-frame:
+
+```python
+# At init time -- done once
+for c in all_characters:
+    img = Image.new("L", (cell_w, cell_h), 0)
+    ImageDraw.Draw(img).text((0, 0), c, fill=255, font=font)
+    bitmaps[c] = np.array(img, dtype=np.float32) / 255.0  # float32 for fast multiply
+
+# At render time -- fast lookup
+bitmap = bitmaps[char]
+canvas[y:y+ch, x:x+cw] = np.maximum(canvas[y:y+ch, x:x+cw],
+                                      (bitmap[:,:,None] * color).astype(np.uint8))
+```
+
+Collect all characters from all palettes + overlay text into the init set. Lazy-init for any missed characters.
+
+## Coordinate Array Caching
+
+Pre-compute all grid-relative coordinate arrays at init, not per-frame:
+
+```python
+# These are O(rows*cols) and used in every effect
+self.rr = np.arange(rows)[:, None]    # row indices
+self.cc = np.arange(cols)[None, :]    # col indices
+self.dist = np.sqrt(dx**2 + dy**2)   # distance from center
+self.angle = np.arctan2(dy, dx)       # angle from center
+self.dist_n = ...                      # normalized distance
+```
+
+## Vectorized Effect Patterns
+
+### Avoid Per-Cell Python Loops in Effects
+
+The render loop (compositing bitmaps) is unavoidably per-cell. But effect functions must be fully vectorized numpy -- never iterate over rows/cols in Python.
+
+Bad (O(rows*cols) Python loop):
+```python
+for r in range(rows):
+    for c in range(cols):
+        val[r, c] = math.sin(c * 0.1 + t) * math.cos(r * 0.1 - t)
+```
+
+Good (vectorized):
+```python
+val = np.sin(g.cc * 0.1 + t) * np.cos(g.rr * 0.1 - t)
+```
+
+### Vectorized Matrix Rain
+
+The naive per-column per-trail-pixel loop is the second biggest bottleneck after the render loop. Use numpy fancy indexing:
+
+```python
+# Instead of nested Python loops over columns and trail pixels:
+# Build row index arrays for all active trail pixels at once
+all_rows = []
+all_cols = []
+all_fades = []
+for c in range(cols):
+    head = int(state["ry"][c])
+    trail_len = state["rln"][c]
+    for i in range(trail_len):
+        row = head - i
+        if 0 <= row < rows:
+            all_rows.append(row)
+            all_cols.append(c)
+            all_fades.append(1.0 - i / trail_len)
+
+# Vectorized assignment
+ar = np.array(all_rows)
+ac = np.array(all_cols)
+af = np.array(all_fades, dtype=np.float32)
+# Assign chars and colors in bulk using fancy indexing
+ch[ar, ac] = ...  # vectorized char assignment
+co[ar, ac, 1] = (af * bri * 255).astype(np.uint8)  # green channel
+```
+
+### Vectorized Fire Columns
+
+Same pattern -- accumulate index arrays, assign in bulk:
+
+```python
+fire_val = np.zeros((rows, cols), dtype=np.float32)
+for fi in range(n_cols):
+    fx_c = int((fi * cols / n_cols + np.sin(t * 2 + fi * 0.7) * 3) % cols)
+    height = int(energy * rows * 0.7)
+    dy = np.arange(min(height, rows))
+    fr = rows - 1 - dy
+    frac = dy / max(height, 1)
+    # Width spread: base columns wider at bottom
+    for dx in range(-1, 2):  # 3-wide columns
+        c = fx_c + dx
+        if 0 <= c < cols:
+            fire_val[fr, c] = np.maximum(fire_val[fr, c],
+                                          (1 - frac * 0.6) * (0.5 + rms * 0.5))
+# Now map fire_val to chars and colors in one vectorized pass
+```
+
+## Bloom Optimization
+
+**Do NOT use `scipy.ndimage.uniform_filter`** -- measured at 424ms/frame.
+
+Use 4x downsample + manual box blur instead -- 84ms/frame (5x faster):
+
+```python
+sm = canvas[::4, ::4].astype(np.float32)  # 4x downsample
+br = np.where(sm > threshold, sm, 0)
+for _ in range(3):                          # 3-pass manual box blur
+    p = np.pad(br, ((1,1),(1,1),(0,0)), mode='edge')
+    br = (p[:-2,:-2] + p[:-2,1:-1] + p[:-2,2:] +
+          p[1:-1,:-2] + p[1:-1,1:-1] + p[1:-1,2:] +
+          p[2:,:-2] + p[2:,1:-1] + p[2:,2:]) / 9.0
+bl = np.repeat(np.repeat(br, 4, axis=0), 4, axis=1)[:H, :W]
+```
+
+## Vignette Caching
+
+Distance field is resolution- and strength-dependent, never changes per frame:
+
+```python
+_vig_cache = {}
+def sh_vignette(canvas, strength):
+    key = (canvas.shape[0], canvas.shape[1], round(strength, 2))
+    if key not in _vig_cache:
+        Y = np.linspace(-1, 1, H)[:, None]
+        X = np.linspace(-1, 1, W)[None, :]
+        _vig_cache[key] = np.clip(1.0 - np.sqrt(X**2+Y**2) * strength, 0.15, 1).astype(np.float32)
+    return np.clip(canvas * _vig_cache[key][:,:,None], 0, 255).astype(np.uint8)
+```
+
+Same pattern for CRT barrel distortion (cache remap coordinates).
+
+## Film Grain Optimization
+
+Generate noise at half resolution, tile up:
+
+```python
+noise = np.random.randint(-amt, amt+1, (H//2, W//2, 1), dtype=np.int16)
+noise = np.repeat(np.repeat(noise, 2, axis=0), 2, axis=1)[:H, :W]
+```
+
+2x blocky grain looks like film grain and costs 1/4 the random generation.
+
+## Parallel Rendering
+
+### Worker Architecture
+
+```python
+hw = detect_hardware()
+N_WORKERS = hw["workers"]
+
+# Batch splitting (for non-clip architectures)
+batch_size = (n_frames + N_WORKERS - 1) // N_WORKERS
+batches = [(i, i*batch_size, min((i+1)*batch_size, n_frames), features, seg_path) ...]
+
+with multiprocessing.Pool(N_WORKERS) as pool:
+    segments = pool.starmap(render_batch, batches)
+```
+
+### Per-Clip Parallelism (Preferred for Segmented Videos)
+
+```python
+from concurrent.futures import ProcessPoolExecutor, as_completed
+
+with ProcessPoolExecutor(max_workers=N_WORKERS) as pool:
+    futures = {pool.submit(render_clip, seg, features, path): seg["id"]
+               for seg, path in clip_args}
+    for fut in as_completed(futures):
+        clip_id = futures[fut]
+        try:
+            fut.result()
+            log(f"  {clip_id} done")
+        except Exception as e:
+            log(f"  {clip_id} FAILED: {e}")
+```
+
+### Worker Isolation
+
+Each worker:
+- Creates its own `Renderer` instance (with full grid + bitmap init)
+- Opens its own ffmpeg subprocess
+- Has independent random seed (`random.seed(batch_id * 10000)`)
+- Writes to its own segment file and stderr log
+
+### ffmpeg Pipe Safety
+
+**CRITICAL**: Never `stderr=subprocess.PIPE` with long-running ffmpeg. The stderr buffer fills at ~64KB and deadlocks:
+
+```python
+# WRONG -- will deadlock
+pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE, stderr=subprocess.PIPE)
+
+# RIGHT -- stderr to file
+stderr_fh = open(err_path, "w")
+pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.DEVNULL, stderr=stderr_fh)
+# ... write all frames ...
+pipe.stdin.close()
+pipe.wait()
+stderr_fh.close()
+```
+
+### Concatenation
+
+```python
+with open(concat_file, "w") as cf:
+    for seg in segments:
+        cf.write(f"file '{seg}'\n")
+
+cmd = ["ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", concat_file]
+if audio_path:
+    cmd += ["-i", audio_path, "-c:v", "copy", "-c:a", "aac", "-b:a", "192k", "-shortest"]
+else:
+    cmd += ["-c:v", "copy"]
+cmd.append(output_path)
+subprocess.run(cmd, capture_output=True, check=True)
+```
+
+## Particle System Performance
+
+Cap particle counts based on quality profile:
+
+| System | Low | Standard | High |
+|--------|-----|----------|------|
+| Explosion | 300 | 1000 | 2500 |
+| Embers | 500 | 1500 | 3000 |
+| Starfield | 300 | 800 | 1500 |
+| Dissolve | 200 | 600 | 1200 |
+
+Cull by truncating lists:
+```python
+MAX_PARTICLES = profile.get("particles_max", 1200)
+if len(S["px"]) > MAX_PARTICLES:
+    for k in ("px", "py", "vx", "vy", "life", "char"):
+        S[k] = S[k][-MAX_PARTICLES:]  # keep newest
+```
+
+## Memory Management
+
+- Feature arrays: pre-computed for all frames, shared across workers via fork semantics (COW)
+- Canvas: allocated once per worker, reused (`np.zeros(...)`)
+- Character arrays: allocated per frame (cheap -- rows*cols U1 strings)
+- Bitmap cache: ~500KB per grid size, initialized once per worker
+
+Total memory per worker: ~50-150MB. Total: ~400-800MB for 8 workers.
+
+For low-memory systems (< 4GB), reduce worker count and use smaller grids.
+
+## Brightness Verification
+
+After render, spot-check brightness at sample timestamps:
+
+```python
+for t in [2, 30, 60, 120, 180]:
+    cmd = ["ffmpeg", "-ss", str(t), "-i", output_path,
+           "-frames:v", "1", "-f", "rawvideo", "-pix_fmt", "rgb24", "-"]
+    r = subprocess.run(cmd, capture_output=True)
+    arr = np.frombuffer(r.stdout, dtype=np.uint8)
+    print(f"t={t}s  mean={arr.mean():.1f}  max={arr.max()}")
+```
+
+Target: mean > 5 for quiet sections, mean > 15 for active sections. If consistently below, increase brightness floor in effects and/or global boost multiplier.
+
+## Render Time Estimates
+
+Scale with hardware. Baseline: 1080p, 24fps, ~180ms/frame/worker.
+
+| Duration | Frames | 4 workers | 8 workers | 16 workers |
+|----------|--------|-----------|-----------|------------|
+| 30s | 720 | ~3 min | ~2 min | ~1 min |
+| 2 min | 2,880 | ~13 min | ~7 min | ~4 min |
+| 3.5 min | 5,040 | ~23 min | ~12 min | ~6 min |
+| 5 min | 7,200 | ~33 min | ~17 min | ~9 min |
+| 10 min | 14,400 | ~65 min | ~33 min | ~17 min |
+
+At 720p: multiply times by ~0.5. At 4K: multiply by ~4.
+
+Heavier effects (many particles, dense grids, extra shader passes) add ~20-50%.
--- a/skills/creative/ascii-video/references/scenes.md
+++ b/skills/creative/ascii-video/references/scenes.md
@@ -0,0 +1,382 @@
+# Scene System Reference
+
+Scenes are the top-level creative unit. Each scene is a time-bounded segment with its own effect function, shader chain, feedback configuration, and tone-mapping gamma.
+
+## Scene Protocol (v2)
+
+### Function Signature
+
+```python
+def fx_scene_name(r, f, t, S) -> canvas:
+    """
+    Args:
+        r: Renderer instance — access multiple grids via r.get_grid("sm")
+        f: dict of audio/video features, all values normalized to [0, 1]
+        t: time in seconds (global, not local to scene)
+        S: dict for persistent state (particles, rain columns, etc.)
+
+    Returns:
+        canvas: numpy uint8 array, shape (VH, VW, 3) — full pixel frame
+    """
+```
+
+This replaces the v1 protocol where scenes returned `(chars, colors)` tuples. The v2 protocol gives scenes full control over multi-grid rendering and pixel-level composition internally.
+
+### The Renderer Class
+
+```python
+class Renderer:
+    def __init__(self):
+        self.grids = {}   # lazy-initialized grid cache
+        self.g = None      # "active" grid (for backward compat)
+        self.S = {}        # persistent state dict
+
+    def get_grid(self, key):
+        """Get or create a GridLayer by size key."""
+        if key not in self.grids:
+            sizes = {"xs": 8, "sm": 10, "md": 16, "lg": 20, "xl": 24, "xxl": 40}
+            self.grids[key] = GridLayer(FONT_PATH, sizes[key])
+        return self.grids[key]
+
+    def set_grid(self, key):
+        """Set active grid (legacy). Prefer get_grid() for multi-grid scenes."""
+        self.g = self.get_grid(key)
+        return self.g
+```
+
+**Key difference from v1**: scenes call `r.get_grid("sm")`, `r.get_grid("lg")`, etc. to access multiple grids. Each grid is lazy-initialized and cached. The `set_grid()` method still works for single-grid scenes.
+
+### Minimal Scene (Single Grid)
+
+```python
+def fx_simple_rings(r, f, t, S):
+    """Single-grid scene: rings with distance-mapped hue."""
+    canvas = _render_vf(r, "md",
+        lambda g, f, t, S: vf_rings(g, f, t, S, n_base=8, spacing_base=3),
+        hf_distance(0.3, 0.02), PAL_STARS, f, t, S, sat=0.85)
+    return canvas
+```
+
+### Standard Scene (Two Grids + Blend)
+
+```python
+def fx_tunnel_ripple(r, f, t, S):
+    """Two-grid scene: tunnel depth exclusion-blended with ripple."""
+    canvas_a = _render_vf(r, "md",
+        lambda g, f, t, S: vf_tunnel(g, f, t, S, speed=5.0, complexity=10) * 1.3,
+        hf_distance(0.55, 0.02), PAL_GREEK, f, t, S, sat=0.7)
+
+    canvas_b = _render_vf(r, "sm",
+        lambda g, f, t, S: vf_ripple(g, f, t, S,
+            sources=[(0.3,0.3), (0.7,0.7), (0.5,0.2)], freq=0.5, damping=0.012) * 1.4,
+        hf_angle(0.1), PAL_STARS, f, t, S, sat=0.8)
+
+    return blend_canvas(canvas_a, canvas_b, "exclusion", 0.8)
+```
+
+### Complex Scene (Three Grids + Conditional + Custom Rendering)
+
+```python
+def fx_rings_explosion(r, f, t, S):
+    """Three-grid scene with particles and conditional kaleidoscope."""
+    # Layer 1: rings
+    canvas_a = _render_vf(r, "sm",
+        lambda g, f, t, S: vf_rings(g, f, t, S, n_base=10, spacing_base=2) * 1.4,
+        lambda g, f, t, S: (g.angle / (2*np.pi) + t * 0.15) % 1.0,
+        PAL_STARS, f, t, S, sat=0.9)
+
+    # Layer 2: vortex on different grid
+    canvas_b = _render_vf(r, "md",
+        lambda g, f, t, S: vf_vortex(g, f, t, S, twist=6.0) * 1.2,
+        hf_time_cycle(0.15), PAL_BLOCKS, f, t, S, sat=0.8)
+
+    result = blend_canvas(canvas_b, canvas_a, "screen", 0.7)
+
+    # Layer 3: particles (custom rendering, not _render_vf)
+    g = r.get_grid("sm")
+    if "px" not in S:
+        S["px"], S["py"], S["vx"], S["vy"], S["life"], S["pch"] = (
+            [], [], [], [], [], [])
+    if f.get("beat", 0) > 0.5:
+        chars = list("\u2605\u2736\u2733\u2738\u2726\u2728*+")
+        for _ in range(int(80 + f.get("rms", 0.3) * 120)):
+            ang = random.uniform(0, 2 * math.pi)
+            sp = random.uniform(1, 10) * (0.5 + f.get("sub_r", 0.3) * 2)
+            S["px"].append(float(g.cols // 2))
+            S["py"].append(float(g.rows // 2))
+            S["vx"].append(math.cos(ang) * sp * 2.5)
+            S["vy"].append(math.sin(ang) * sp)
+            S["life"].append(1.0)
+            S["pch"].append(random.choice(chars))
+
+    # Update + draw particles
+    ch_p = np.full((g.rows, g.cols), " ", dtype="U1")
+    co_p = np.zeros((g.rows, g.cols, 3), dtype=np.uint8)
+    i = 0
+    while i < len(S["px"]):
+        S["px"][i] += S["vx"][i]; S["py"][i] += S["vy"][i]
+        S["vy"][i] += 0.03; S["life"][i] -= 0.02
+        if S["life"][i] <= 0:
+            for k in ("px","py","vx","vy","life","pch"): S[k].pop(i)
+        else:
+            pr, pc = int(S["py"][i]), int(S["px"][i])
+            if 0 <= pr < g.rows and 0 <= pc < g.cols:
+                ch_p[pr, pc] = S["pch"][i]
+                co_p[pr, pc] = hsv2rgb_scalar(
+                    0.08 + (1-S["life"][i])*0.15, 0.95, S["life"][i])
+            i += 1
+
+    canvas_p = g.render(ch_p, co_p)
+    result = blend_canvas(result, canvas_p, "add", 0.8)
+
+    # Conditional kaleidoscope on strong beats
+    if f.get("bdecay", 0) > 0.4:
+        result = sh_kaleidoscope(result.copy(), folds=6)
+
+    return result
+```
+
+### Scene with Custom Character Rendering (Matrix Rain)
+
+When you need per-cell control beyond what `_render_vf()` provides:
+
+```python
+def fx_matrix_layered(r, f, t, S):
+    """Matrix rain blended with tunnel — two grids, screen blend."""
+    # Layer 1: Matrix rain (custom per-column rendering)
+    g = r.get_grid("md")
+    rows, cols = g.rows, g.cols
+    pal = PAL_KATA
+
+    if "ry" not in S or len(S["ry"]) != cols:
+        S["ry"] = np.random.uniform(-rows, rows, cols).astype(np.float32)
+        S["rsp"] = np.random.uniform(0.3, 2.0, cols).astype(np.float32)
+        S["rln"] = np.random.randint(8, 35, cols)
+        S["rch"] = np.random.randint(1, len(pal), (rows, cols))
+
+    speed = 0.6 + f.get("bass", 0.3) * 3
+    if f.get("beat", 0) > 0.5: speed *= 2.5
+    S["ry"] += S["rsp"] * speed
+
+    ch = np.full((rows, cols), " ", dtype="U1")
+    co = np.zeros((rows, cols, 3), dtype=np.uint8)
+    heads = S["ry"].astype(int)
+    for c in range(cols):
+        head = heads[c]
+        for i in range(S["rln"][c]):
+            row = head - i
+            if 0 <= row < rows:
+                fade = 1.0 - i / S["rln"][c]
+                ch[row, c] = pal[S["rch"][row, c] % len(pal)]
+                if i == 0:
+                    v = int(min(255, fade * 300))
+                    co[row, c] = (int(v*0.9), v, int(v*0.9))
+                else:
+                    v = int(fade * 240)
+                    co[row, c] = (int(v*0.1), v, int(v*0.4))
+    canvas_a = g.render(ch, co)
+
+    # Layer 2: Tunnel on sm grid for depth texture
+    canvas_b = _render_vf(r, "sm",
+        lambda g, f, t, S: vf_tunnel(g, f, t, S, speed=5.0, complexity=10),
+        hf_distance(0.3, 0.02), PAL_BLOCKS, f, t, S, sat=0.6)
+
+    return blend_canvas(canvas_a, canvas_b, "screen", 0.5)
+```
+
+---
+
+## Scene Table
+
+The scene table defines the timeline: which scene plays when, with what configuration.
+
+### Structure
+
+```python
+SCENES = [
+    {
+        "start": 0.0,           # start time in seconds
+        "end": 3.96,            # end time in seconds
+        "name": "starfield",    # identifier (used for clip filenames)
+        "grid": "sm",           # default grid (for render_clip setup)
+        "fx": fx_starfield,     # scene function reference (must be module-level)
+        "gamma": 0.75,          # tonemap gamma override (default 0.75)
+        "shaders": [            # shader chain (applied after tonemap + feedback)
+            ("bloom", {"thr": 120}),
+            ("vignette", {"s": 0.2}),
+            ("grain", {"amt": 8}),
+        ],
+        "feedback": None,       # feedback buffer config (None = disabled)
+        # "feedback": {"decay": 0.8, "blend": "screen", "opacity": 0.3,
+        #              "transform": "zoom", "transform_amt": 0.02, "hue_shift": 0.02},
+    },
+    {
+        "start": 3.96,
+        "end": 6.58,
+        "name": "matrix_layered",
+        "grid": "md",
+        "fx": fx_matrix_layered,
+        "shaders": [
+            ("crt", {"strength": 0.05}),
+            ("scanlines", {"intensity": 0.12}),
+            ("color_grade", {"tint": (0.7, 1.2, 0.7)}),
+            ("bloom", {"thr": 100}),
+        ],
+        "feedback": {"decay": 0.5, "blend": "add", "opacity": 0.2},
+    },
+    # ... more scenes ...
+]
+```
+
+### Beat-Synced Scene Cutting
+
+Derive cut points from audio analysis:
+
+```python
+# Get beat timestamps
+beats = [fi / FPS for fi in range(N_FRAMES) if features["beat"][fi] > 0.5]
+
+# Group beats into phrase boundaries (every 4-8 beats)
+cuts = [0.0]
+for i in range(0, len(beats), 4):  # cut every 4 beats
+    cuts.append(beats[i])
+cuts.append(DURATION)
+
+# Or use the music's structure: silence gaps, energy changes
+energy = features["rms"]
+# Find timestamps where energy drops significantly -> natural break points
+```
+
+### `render_clip()` — The Render Loop
+
+This function renders one scene to a clip file:
+
+```python
+def render_clip(seg, features, clip_path):
+    r = Renderer()
+    r.set_grid(seg["grid"])
+    S = r.S
+    random.seed(hash(seg["id"]) + 42)  # deterministic per scene
+
+    # Build shader chain from config
+    chain = ShaderChain()
+    for shader_name, kwargs in seg.get("shaders", []):
+        chain.add(shader_name, **kwargs)
+
+    # Setup feedback buffer
+    fb = None
+    fb_cfg = seg.get("feedback", None)
+    if fb_cfg:
+        fb = FeedbackBuffer()
+
+    fx_fn = seg["fx"]
+
+    # Open ffmpeg pipe
+    cmd = ["ffmpeg", "-y", "-f", "rawvideo", "-pix_fmt", "rgb24",
+           "-s", f"{VW}x{VH}", "-r", str(FPS), "-i", "pipe:0",
+           "-c:v", "libx264", "-preset", "fast", "-crf", "20",
+           "-pix_fmt", "yuv420p", clip_path]
+    stderr_fh = open(clip_path.replace(".mp4", ".log"), "w")
+    pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE,
+                            stdout=subprocess.DEVNULL, stderr=stderr_fh)
+
+    for fi in range(seg["frame_start"], seg["frame_end"]):
+        t = fi / FPS
+        feat = {k: float(features[k][fi]) for k in features}
+
+        # 1. Scene renders canvas
+        canvas = fx_fn(r, feat, t, S)
+
+        # 2. Tonemap normalizes brightness
+        canvas = tonemap(canvas, gamma=seg.get("gamma", 0.75))
+
+        # 3. Feedback adds temporal recursion
+        if fb and fb_cfg:
+            canvas = fb.apply(canvas, **{k: fb_cfg[k] for k in fb_cfg})
+
+        # 4. Shader chain adds post-processing
+        canvas = chain.apply(canvas, f=feat, t=t)
+
+        pipe.stdin.write(canvas.tobytes())
+
+    pipe.stdin.close(); pipe.wait(); stderr_fh.close()
+```
+
+### Building Segments from Scene Table
+
+```python
+segments = []
+for i, scene in enumerate(SCENES):
+    segments.append({
+        "id": f"s{i:02d}_{scene['name']}",
+        "name": scene["name"],
+        "grid": scene["grid"],
+        "fx": scene["fx"],
+        "shaders": scene.get("shaders", []),
+        "feedback": scene.get("feedback", None),
+        "gamma": scene.get("gamma", 0.75),
+        "frame_start": int(scene["start"] * FPS),
+        "frame_end": int(scene["end"] * FPS),
+    })
+```
+
+### Parallel Rendering
+
+Scenes are independent units dispatched to a process pool:
+
+```python
+from concurrent.futures import ProcessPoolExecutor, as_completed
+
+with ProcessPoolExecutor(max_workers=N_WORKERS) as pool:
+    futures = {
+        pool.submit(render_clip, seg, features, clip_path): seg["id"]
+        for seg, clip_path in zip(segments, clip_paths)
+    }
+    for fut in as_completed(futures):
+        try:
+            fut.result()
+        except Exception as e:
+            log(f"ERROR {futures[fut]}: {e}")
+```
+
+**Pickling constraint**: `ProcessPoolExecutor` serializes arguments via pickle. Module-level functions can be pickled; lambdas and closures cannot. All `fx_*` scene functions MUST be defined at module level, not as closures or class methods.
+
+### Test-Frame Mode
+
+Render a single frame at a specific timestamp to verify visuals without a full render:
+
+```python
+if args.test_frame >= 0:
+    fi = min(int(args.test_frame * FPS), N_FRAMES - 1)
+    t = fi / FPS
+    feat = {k: float(features[k][fi]) for k in features}
+    scene = next(sc for sc in reversed(SCENES) if t >= sc["start"])
+    r = Renderer()
+    r.set_grid(scene["grid"])
+    canvas = scene["fx"](r, feat, t, r.S)
+    canvas = tonemap(canvas, gamma=scene.get("gamma", 0.75))
+    chain = ShaderChain()
+    for sn, kw in scene.get("shaders", []):
+        chain.add(sn, **kw)
+    canvas = chain.apply(canvas, f=feat, t=t)
+    Image.fromarray(canvas).save(f"test_{args.test_frame:.1f}s.png")
+    print(f"Mean brightness: {canvas.astype(float).mean():.1f}")
+```
+
+CLI: `python reel.py --test-frame 10.0`
+
+---
+
+## Scene Design Checklist
+
+For each scene:
+
+1. **Choose 2-3 grid sizes** — different scales create interference
+2. **Choose different value fields** per layer — don't use the same effect on every grid
+3. **Choose different hue fields** per layer — or at minimum different hue offsets
+4. **Choose different palettes** per layer — mixing PAL_RUNE with PAL_BLOCKS looks different from PAL_RUNE with PAL_DENSE
+5. **Choose a blend mode** that matches the energy — screen for bright, difference for psychedelic, exclusion for subtle
+6. **Add conditional effects** on beat — kaleidoscope, mirror, glitch
+7. **Configure feedback** for trailing/recursive looks — or None for clean cuts
+8. **Set gamma** if using destructive shaders (solarize, posterize)
+9. **Test with --test-frame** at the scene's midpoint before full render
--- a/skills/creative/ascii-video/references/shaders.md
+++ b/skills/creative/ascii-video/references/shaders.md
--- a/skills/creative/ascii-video/references/troubleshooting.md
+++ b/skills/creative/ascii-video/references/troubleshooting.md
@@ -0,0 +1,331 @@
+# Troubleshooting Reference
+
+Common bugs, gotchas, and platform-specific issues encountered during ASCII video development.
+
+## NumPy Broadcasting
+
+### The `broadcast_to().copy()` Trap
+
+Hue field generators often return arrays that are broadcast views — they have shape `(1, cols)` or `(rows, 1)` that numpy broadcasts to `(rows, cols)`. These views are **read-only**. If any downstream code tries to modify them in-place (e.g., `h %= 1.0`), numpy raises:
+
+```
+ValueError: output array is read-only
+```
+
+**Fix**: Always `.copy()` after `broadcast_to()`:
+
+```python
+h = np.broadcast_to(h, (g.rows, g.cols)).copy()
+```
+
+This is especially important in `_render_vf()` where hue arrays flow through `hsv2rgb()`.
+
+### The `+=` vs `+` Trap
+
+Broadcasting also fails with in-place operators when operand shapes don't match exactly:
+
+```python
+# FAILS if result is (rows,1) and operand is (rows, cols)
+val += np.sin(g.cc * 0.02 + t * 0.3) * 0.5
+
+# WORKS — creates a new array
+val = val + np.sin(g.cc * 0.02 + t * 0.3) * 0.5
+```
+
+The `vf_plasma()` function had this bug. Use `+` instead of `+=` when mixing different-shaped arrays.
+
+### Shape Mismatch in `hsv2rgb()`
+
+`hsv2rgb(h, s, v)` requires all three arrays to have identical shapes. If `h` is `(1, cols)` and `s` is `(rows, cols)`, the function crashes or produces wrong output.
+
+**Fix**: Ensure all inputs are broadcast and copied to `(rows, cols)` before calling.
+
+---
+
+## Blend Mode Pitfalls
+
+### Overlay Crushes Dark Inputs
+
+`overlay(a, b) = 2*a*b` when `a < 0.5`. Two values of 0.12 produce `2 * 0.12 * 0.12 = 0.03`. The result is darker than either input.
+
+**Impact**: If both layers are dark (which ASCII art usually is), overlay produces near-black output.
+
+**Fix**: Use `screen` for dark source material. Screen always brightens: `1 - (1-a)*(1-b)`.
+
+### Colordodge Division by Zero
+
+`colordodge(a, b) = a / (1 - b)`. When `b = 1.0` (pure white pixels), this divides by zero.
+
+**Fix**: Add epsilon: `a / (1 - b + 1e-6)`. The implementation in `BLEND_MODES` should include this.
+
+### Colorburn Division by Zero
+
+`colorburn(a, b) = 1 - (1-a) / b`. When `b = 0` (pure black pixels), this divides by zero.
+
+**Fix**: Add epsilon: `1 - (1-a) / (b + 1e-6)`.
+
+### Multiply Always Darkens
+
+`multiply(a, b) = a * b`. Since both operands are [0,1], the result is always <= min(a,b). Never use multiply as a feedback blend mode — the frame goes black within a few frames.
+
+**Fix**: Use `screen` for feedback, or `add` with low opacity.
+
+---
+
+## Multiprocessing
+
+### Pickling Constraints
+
+`ProcessPoolExecutor` serializes function arguments via pickle. This constrains what you can pass to workers:
+
+| Can Pickle | Cannot Pickle |
+|-----------|---------------|
+| Module-level functions (`def fx_foo():`) | Lambdas (`lambda x: x + 1`) |
+| Dicts, lists, numpy arrays | Closures (functions defined inside functions) |
+| Class instances (with `__reduce__`) | Instance methods |
+| Strings, numbers | File handles, sockets |
+
+**Impact**: All scene functions referenced in the SCENES table must be defined at module level with `def`. If you use a lambda or closure, you get:
+
+```
+_pickle.PicklingError: Can't pickle <function <lambda> at 0x...>
+```
+
+**Fix**: Define all scene functions at module top level. Lambdas used inside `_render_vf()` as val_fn/hue_fn are fine because they execute within the worker process — they're not pickled across process boundaries.
+
+### macOS spawn vs Linux fork
+
+On macOS, `multiprocessing` defaults to `spawn` (full serialization). On Linux, it defaults to `fork` (copy-on-write). This means:
+
+- **macOS**: Feature arrays are serialized per worker (~57KB for 30s video, but scales with duration). Each worker re-imports the entire module.
+- **Linux**: Feature arrays are shared via COW. Workers inherit the parent's memory.
+
+**Impact**: On macOS, module-level code (like `detect_hardware()`) runs in every worker process. If it has side effects (e.g., subprocess calls), those happen N+1 times.
+
+### Per-Worker State Isolation
+
+Each worker creates its own:
+- `Renderer` instance (with fresh grid cache)
+- `FeedbackBuffer` (feedback doesn't cross scene boundaries)
+- Random seed (`random.seed(hash(seg_id) + 42)`)
+
+This means:
+- Particle state doesn't carry between scenes (expected)
+- Feedback trails reset at scene cuts (expected)
+- `np.random` state is NOT seeded by `random.seed()` — they use separate RNGs
+
+**Fix for deterministic noise**: Use `np.random.RandomState(seed)` explicitly:
+
+```python
+rng = np.random.RandomState(hash(seg_id) + 42)
+noise = rng.random((rows, cols))
+```
+
+---
+
+## Brightness Issues
+
+### Dark Scenes After Tonemap
+
+If a scene is still dark after tonemap, check:
+
+1. **Gamma too high**: Lower gamma (0.5-0.6) for scenes with destructive post-processing
+2. **Shader destroying brightness**: Solarize, posterize, or contrast adjustments in the shader chain can undo tonemap's work. Move destructive shaders earlier in the chain, or increase gamma to compensate.
+3. **Feedback with multiply**: Multiply feedback darkens every frame. Switch to screen or add.
+4. **Overlay blend in scene**: If the scene function uses `blend_canvas(..., "overlay", ...)` with dark layers, switch to screen.
+
+### Diagnostic: Test-Frame Brightness
+
+```bash
+python reel.py --test-frame 10.0
+# Output: Mean brightness: 44.3, max: 255
+```
+
+If mean < 20, the scene needs attention. Common fixes:
+- Lower gamma in the SCENES entry
+- Change internal blend modes from overlay/multiply to screen/add
+- Increase value field multipliers (e.g., `vf_plasma(...) * 1.5`)
+- Check that the shader chain doesn't have an aggressive solarize or threshold
+
+### v1 Brightness Pattern (Deprecated)
+
+The old pattern used a linear multiplier:
+
+```python
+# OLD — don't use
+canvas = np.clip(canvas.astype(np.float32) * 2.0, 0, 255).astype(np.uint8)
+```
+
+This fails because:
+- Dark scenes (mean 8): `8 * 2.0 = 16` — still dark
+- Bright scenes (mean 130): `130 * 2.0 = 255` — clipped, lost detail
+
+Use `tonemap()` instead. See `composition.md` § Adaptive Tone Mapping.
+
+---
+
+## ffmpeg Issues
+
+### Pipe Deadlock
+
+The #1 production bug. If you use `stderr=subprocess.PIPE`:
+
+```python
+# DEADLOCK — stderr buffer fills at 64KB, blocks ffmpeg, blocks your writes
+pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE, stderr=subprocess.PIPE)
+```
+
+**Fix**: Always redirect stderr to a file:
+
+```python
+stderr_fh = open(err_path, "w")
+pipe = subprocess.Popen(cmd, stdin=subprocess.PIPE,
+                        stdout=subprocess.DEVNULL, stderr=stderr_fh)
+```
+
+### Frame Count Mismatch
+
+If the number of frames written to the pipe doesn't match what ffmpeg expects (based on `-r` and duration), the output may have:
+- Missing frames at the end
+- Incorrect duration
+- Audio-video desync
+
+**Fix**: Calculate frame count explicitly: `n_frames = int(duration * FPS)`. Don't use `range(int(start*FPS), int(end*FPS))` without verifying the total matches.
+
+### Concat Fails with "unsafe file name"
+
+```
+[concat @ ...] Unsafe file name
+```
+
+**Fix**: Always use `-safe 0`:
+```python
+["ffmpeg", "-f", "concat", "-safe", "0", "-i", concat_path, ...]
+```
+
+---
+
+## Font Issues
+
+### Cell Height (macOS Pillow)
+
+`textbbox()` and `getbbox()` return incorrect heights on some macOS Pillow versions. Use `getmetrics()`:
+
+```python
+ascent, descent = font.getmetrics()
+cell_height = ascent + descent  # correct
+# NOT: font.getbbox("M")[3]  # wrong on some versions
+```
+
+### Missing Unicode Glyphs
+
+Not all fonts render all Unicode characters. If a palette character isn't in the font, the glyph renders as a blank or tofu box, appearing as a dark hole in the output.
+
+**Fix**: Validate at init:
+
+```python
+all_chars = set()
+for pal in [PAL_DEFAULT, PAL_DENSE, PAL_RUNE, ...]:
+    all_chars.update(pal)
+
+valid_chars = set()
+for c in all_chars:
+    if c == " ":
+        valid_chars.add(c)
+        continue
+    img = Image.new("L", (20, 20), 0)
+    ImageDraw.Draw(img).text((0, 0), c, fill=255, font=font)
+    if np.array(img).max() > 0:
+        valid_chars.add(c)
+    else:
+        log(f"WARNING: '{c}' (U+{ord(c):04X}) missing from font")
+```
+
+### Platform Font Paths
+
+| Platform | Common Paths |
+|----------|-------------|
+| macOS | `/System/Library/Fonts/Menlo.ttc`, `/System/Library/Fonts/Monaco.ttf` |
+| Linux | `/usr/share/fonts/truetype/dejavu/DejaVuSansMono.ttf` |
+| Windows | `C:\Windows\Fonts\consola.ttf` (Consolas) |
+
+Always probe multiple paths and fall back gracefully. See `architecture.md` § Font Selection.
+
+---
+
+## Performance
+
+### Slow Shaders
+
+Some shaders use Python loops and are very slow at 1080p:
+
+| Shader | Issue | Fix |
+|--------|-------|-----|
+| `wave_distort` | Per-row Python loop | Use vectorized fancy indexing |
+| `halftone` | Triple-nested loop | Vectorize with block reduction |
+| `matrix rain` | Per-column per-trail loop | Accumulate index arrays, bulk assign |
+
+### Render Time Scaling
+
+If render is taking much longer than expected:
+1. Check grid count — each extra grid adds ~100-150ms/frame for init
+2. Check particle count — cap at quality-appropriate limits
+3. Check shader count — each shader adds 2-25ms
+4. Check for accidental Python loops in effects (should be numpy only)
+
+---
+
+## Common Mistakes
+
+### Using `r.S` vs the `S` Parameter
+
+The v2 scene protocol passes `S` (the state dict) as an explicit parameter. But `S` IS `r.S` — they're the same object. Both work:
+
+```python
+def fx_scene(r, f, t, S):
+    S["counter"] = S.get("counter", 0) + 1   # via parameter (preferred)
+    r.S["counter"] = r.S.get("counter", 0) + 1  # via renderer (also works)
+```
+
+Use the `S` parameter for clarity. The explicit parameter makes it obvious that the function has persistent state.
+
+### Forgetting to Handle Empty Feature Values
+
+Audio features default to 0.0 if the audio is silent. Use `.get()` with sensible defaults:
+
+```python
+energy = f.get("bass", 0.3)  # default to 0.3, not 0
+```
+
+If you default to 0, effects go blank during silence.
+
+### Writing New Files Instead of Editing Existing State
+
+A common bug in particle systems: creating new arrays every frame instead of updating persistent state.
+
+```python
+# WRONG — particles reset every frame
+S["px"] = []
+for _ in range(100):
+    S["px"].append(random.random())
+
+# RIGHT — only initialize once, update each frame
+if "px" not in S:
+    S["px"] = []
+# ... emit new particles based on beats
+# ... update existing particles
+```
+
+### Not Clipping Value Fields
+
+Value fields should be [0, 1]. If they exceed this range, `val2char()` produces index errors:
+
+```python
+# WRONG — vf_plasma() * 1.5 can exceed 1.0
+val = vf_plasma(g, f, t, S) * 1.5
+
+# RIGHT — clip after scaling
+val = np.clip(vf_plasma(g, f, t, S) * 1.5, 0, 1)
+```
+
+The `_render_vf()` helper clips automatically, but if you're building custom scenes, clip explicitly.
--- a/skills/diagramming/excalidraw/SKILL.md
+++ b/skills/diagramming/excalidraw/SKILL.md
--- a/skills/diagramming/excalidraw/references/colors.md
+++ b/skills/diagramming/excalidraw/references/colors.md
--- a/skills/diagramming/excalidraw/references/dark-mode.md
+++ b/skills/diagramming/excalidraw/references/dark-mode.md
--- a/skills/diagramming/excalidraw/references/examples.md
+++ b/skills/diagramming/excalidraw/references/examples.md
--- a/skills/diagramming/excalidraw/scripts/upload.py
+++ b/skills/diagramming/excalidraw/scripts/upload.py
--- a/skills/dogfood/SKILL.md
+++ b/skills/dogfood/SKILL.md
@@ -0,0 +1,162 @@
+---
+name: dogfood
+description: Systematic exploratory QA testing of web applications — find bugs, capture evidence, and generate structured reports
+version: 1.0.0
+metadata:
+  hermes:
+    tags: [qa, testing, browser, web, dogfood]
+    related_skills: []
+---
+
+# Dogfood: Systematic Web Application QA Testing
+
+## Overview
+
+This skill guides you through systematic exploratory QA testing of web applications using the browser toolset. You will navigate the application, interact with elements, capture evidence of issues, and produce a structured bug report.
+
+## Prerequisites
+
+- Browser toolset must be available (`browser_navigate`, `browser_snapshot`, `browser_click`, `browser_type`, `browser_vision`, `browser_console`, `browser_scroll`, `browser_back`, `browser_press`, `browser_close`)
+- A target URL and testing scope from the user
+
+## Inputs
+
+The user provides:
+1. **Target URL** — the entry point for testing
+2. **Scope** — what areas/features to focus on (or "full site" for comprehensive testing)
+3. **Output directory** (optional) — where to save screenshots and the report (default: `./dogfood-output`)
+
+## Workflow
+
+Follow this 5-phase systematic workflow:
+
+### Phase 1: Plan
+
+1. Create the output directory structure:
+   ```
+   {output_dir}/
+   ├── screenshots/       # Evidence screenshots
+   └── report.md          # Final report (generated in Phase 5)
+   ```
+2. Identify the testing scope based on user input.
+3. Build a rough sitemap by planning which pages and features to test:
+   - Landing/home page
+   - Navigation links (header, footer, sidebar)
+   - Key user flows (sign up, login, search, checkout, etc.)
+   - Forms and interactive elements
+   - Edge cases (empty states, error pages, 404s)
+
+### Phase 2: Explore
+
+For each page or feature in your plan:
+
+1. **Navigate** to the page:
+   ```
+   browser_navigate(url="https://example.com/page")
+   ```
+
+2. **Take a snapshot** to understand the DOM structure:
+   ```
+   browser_snapshot()
+   ```
+
+3. **Check the console** for JavaScript errors:
+   ```
+   browser_console(clear=true)
+   ```
+   Do this after every navigation and after every significant interaction. Silent JS errors are high-value findings.
+
+4. **Take an annotated screenshot** to visually assess the page and identify interactive elements:
+   ```
+   browser_vision(question="Describe the page layout, identify any visual issues, broken elements, or accessibility concerns", annotate=true)
+   ```
+   The `annotate=true` flag overlays numbered `[N]` labels on interactive elements. Each `[N]` maps to ref `@eN` for subsequent browser commands.
+
+5. **Test interactive elements** systematically:
+   - Click buttons and links: `browser_click(ref="@eN")`
+   - Fill forms: `browser_type(ref="@eN", text="test input")`
+   - Test keyboard navigation: `browser_press(key="Tab")`, `browser_press(key="Enter")`
+   - Scroll through content: `browser_scroll(direction="down")`
+   - Test form validation with invalid inputs
+   - Test empty submissions
+
+6. **After each interaction**, check for:
+   - Console errors: `browser_console()`
+   - Visual changes: `browser_vision(question="What changed after the interaction?")`
+   - Expected vs actual behavior
+
+### Phase 3: Collect Evidence
+
+For every issue found:
+
+1. **Take a screenshot** showing the issue:
+   ```
+   browser_vision(question="Capture and describe the issue visible on this page", annotate=false)
+   ```
+   Save the `screenshot_path` from the response — you will reference it in the report.
+
+2. **Record the details**:
+   - URL where the issue occurs
+   - Steps to reproduce
+   - Expected behavior
+   - Actual behavior
+   - Console errors (if any)
+   - Screenshot path
+
+3. **Classify the issue** using the issue taxonomy (see `references/issue-taxonomy.md`):
+   - Severity: Critical / High / Medium / Low
+   - Category: Functional / Visual / Accessibility / Console / UX / Content
+
+### Phase 4: Categorize
+
+1. Review all collected issues.
+2. De-duplicate — merge issues that are the same bug manifesting in different places.
+3. Assign final severity and category to each issue.
+4. Sort by severity (Critical first, then High, Medium, Low).
+5. Count issues by severity and category for the executive summary.
+
+### Phase 5: Report
+
+Generate the final report using the template at `templates/dogfood-report-template.md`.
+
+The report must include:
+1. **Executive summary** with total issue count, breakdown by severity, and testing scope
+2. **Per-issue sections** with:
+   - Issue number and title
+   - Severity and category badges
+   - URL where observed
+   - Description of the issue
+   - Steps to reproduce
+   - Expected vs actual behavior
+   - Screenshot references (use `MEDIA:<screenshot_path>` for inline images)
+   - Console errors if relevant
+3. **Summary table** of all issues
+4. **Testing notes** — what was tested, what was not, any blockers
+
+Save the report to `{output_dir}/report.md`.
+
+## Tools Reference
+
+| Tool | Purpose |
+|------|---------|
+| `browser_navigate` | Go to a URL |
+| `browser_snapshot` | Get DOM text snapshot (accessibility tree) |
+| `browser_click` | Click an element by ref (`@eN`) or text |
+| `browser_type` | Type into an input field |
+| `browser_scroll` | Scroll up/down on the page |
+| `browser_back` | Go back in browser history |
+| `browser_press` | Press a keyboard key |
+| `browser_vision` | Screenshot + AI analysis; use `annotate=true` for element labels |
+| `browser_console` | Get JS console output and errors |
+| `browser_close` | Close the browser session |
+
+## Tips
+
+- **Always check `browser_console()` after navigating and after significant interactions.** Silent JS errors are among the most valuable findings.
+- **Use `annotate=true` with `browser_vision`** when you need to reason about interactive element positions or when the snapshot refs are unclear.
+- **Test with both valid and invalid inputs** — form validation bugs are common.
+- **Scroll through long pages** — content below the fold may have rendering issues.
+- **Test navigation flows** — click through multi-step processes end-to-end.
+- **Check responsive behavior** by noting any layout issues visible in screenshots.
+- **Don't forget edge cases**: empty states, very long text, special characters, rapid clicking.
+- When reporting screenshots to the user, include `MEDIA:<screenshot_path>` so they can see the evidence inline.
--- a/skills/dogfood/references/issue-taxonomy.md
+++ b/skills/dogfood/references/issue-taxonomy.md
@@ -0,0 +1,109 @@
+# Issue Taxonomy
+
+Use this taxonomy to classify issues found during dogfood QA testing.
+
+## Severity Levels
+
+### Critical
+The issue makes a core feature completely unusable or causes data loss.
+
+**Examples:**
+- Application crashes or shows a blank white page
+- Form submission silently loses user data
+- Authentication is completely broken (can't log in at all)
+- Payment flow fails and charges the user without completing the order
+- Security vulnerability (e.g., XSS, exposed credentials in console)
+
+### High
+The issue significantly impairs functionality but a workaround may exist.
+
+**Examples:**
+- A key button does nothing when clicked (but refreshing fixes it)
+- Search returns no results for valid queries
+- Form validation rejects valid input
+- Page loads but critical content is missing or garbled
+- Navigation link leads to a 404 or wrong page
+- Uncaught JavaScript exceptions in the console on core pages
+
+### Medium
+The issue is noticeable and affects user experience but doesn't block core functionality.
+
+**Examples:**
+- Layout is misaligned or overlapping on certain screen sections
+- Images fail to load (broken image icons)
+- Slow performance (visible loading delays > 3 seconds)
+- Form field lacks proper validation feedback (no error message on bad input)
+- Console warnings that suggest deprecated or misconfigured features
+- Inconsistent styling between similar pages
+
+### Low
+Minor polish issues that don't affect functionality.
+
+**Examples:**
+- Typos or grammatical errors in text content
+- Minor spacing or alignment inconsistencies
+- Placeholder text left in production ("Lorem ipsum")
+- Favicon missing
+- Console info/debug messages that shouldn't be in production
+- Subtle color contrast issues that don't fail WCAG requirements
+
+## Categories
+
+### Functional
+Issues where features don't work as expected.
+
+- Buttons/links that don't respond
+- Forms that don't submit or submit incorrectly
+- Broken user flows (can't complete a multi-step process)
+- Incorrect data displayed
+- Features that work partially
+
+### Visual
+Issues with the visual presentation of the page.
+
+- Layout problems (overlapping elements, broken grids)
+- Broken images or missing media
+- Styling inconsistencies
+- Responsive design failures
+- Z-index issues (elements hidden behind others)
+- Text overflow or truncation
+
+### Accessibility
+Issues that prevent or hinder access for users with disabilities.
+
+- Missing alt text on meaningful images
+- Poor color contrast (fails WCAG AA)
+- Elements not reachable via keyboard navigation
+- Missing form labels or ARIA attributes
+- Focus indicators missing or unclear
+- Screen reader incompatible content
+
+### Console
+Issues detected through JavaScript console output.
+
+- Uncaught exceptions and unhandled promise rejections
+- Failed network requests (4xx, 5xx errors in console)
+- Deprecation warnings
+- CORS errors
+- Mixed content warnings (HTTP resources on HTTPS page)
+- Excessive console.log output left from development
+
+### UX (User Experience)
+Issues where functionality works but the experience is poor.
+
+- Confusing navigation or information architecture
+- Missing loading indicators (user doesn't know something is happening)
+- No feedback after user actions (e.g., button click with no visible result)
+- Inconsistent interaction patterns
+- Missing confirmation dialogs for destructive actions
+- Poor error messages that don't help the user recover
+
+### Content
+Issues with the text, media, or information on the page.
+
+- Typos and grammatical errors
+- Placeholder/dummy content in production
+- Outdated information
+- Missing content (empty sections)
+- Broken or dead links to external resources
+- Incorrect or misleading labels
--- a/skills/dogfood/templates/dogfood-report-template.md
+++ b/skills/dogfood/templates/dogfood-report-template.md
@@ -0,0 +1,86 @@
+# Dogfood QA Report
+
+**Target:** {target_url}
+**Date:** {date}
+**Scope:** {scope_description}
+**Tester:** Hermes Agent (automated exploratory QA)
+
+---
+
+## Executive Summary
+
+| Severity | Count |
+|----------|-------|
+| 🔴 Critical | {critical_count} |
+| 🟠 High | {high_count} |
+| 🟡 Medium | {medium_count} |
+| 🔵 Low | {low_count} |
+| **Total** | **{total_count}** |
+
+**Overall Assessment:** {one_sentence_assessment}
+
+---
+
+## Issues
+
+<!-- Repeat this section for each issue found, sorted by severity (Critical first) -->
+
+### Issue #{issue_number}: {issue_title}
+
+| Field | Value |
+|-------|-------|
+| **Severity** | {severity} |
+| **Category** | {category} |
+| **URL** | {url_where_found} |
+
+**Description:**
+{detailed_description_of_the_issue}
+
+**Steps to Reproduce:**
+1. {step_1}
+2. {step_2}
+3. {step_3}
+
+**Expected Behavior:**
+{what_should_happen}
+
+**Actual Behavior:**
+{what_actually_happens}
+
+**Screenshot:**
+MEDIA:{screenshot_path}
+
+**Console Errors** (if applicable):
+```
+{console_error_output}
+```
+
+---
+
+<!-- End of per-issue section -->
+
+## Issues Summary Table
+
+| # | Title | Severity | Category | URL |
+|---|-------|----------|----------|-----|
+| {n} | {title} | {severity} | {category} | {url} |
+
+## Testing Coverage
+
+### Pages Tested
+- {list_of_pages_visited}
+
+### Features Tested
+- {list_of_features_exercised}
+
+### Not Tested / Out of Scope
+- {areas_not_covered_and_why}
+
+### Blockers
+- {any_issues_that_prevented_testing_certain_areas}
+
+---
+
+## Notes
+
+{any_additional_observations_or_recommendations}
--- a/skills/gaming/pokemon-player/SKILL.md
+++ b/skills/gaming/pokemon-player/SKILL.md
@@ -0,0 +1,215 @@
+---
+name: pokemon-player
+description: Play Pokemon games autonomously via headless emulation. Starts a game server, reads structured game state from RAM, makes strategic decisions, and sends button inputs — all from the terminal.
+tags: [gaming, pokemon, emulator, pyboy, gameplay, gameboy]
+---
+# Pokemon Player
+
+Play Pokemon games via headless emulation using the `pokemon-agent` package.
+
+## When to Use
+- User says "play pokemon", "start pokemon", "pokemon game"
+- User asks about Pokemon Red, Blue, Yellow, FireRed, etc.
+- User wants to watch an AI play Pokemon
+- User references a ROM file (.gb, .gbc, .gba)
+
+## Startup Procedure
+
+### 1. First-time setup (clone, venv, install)
+The repo is NousResearch/pokemon-agent on GitHub. Clone it, then
+set up a Python 3.10+ virtual environment. Use uv (preferred for speed)
+to create the venv and install the package in editable mode with the
+pyboy extra. If uv is not available, fall back to python3 -m venv + pip.
+
+On this machine it is already set up at /home/teknium/pokemon-agent
+with a venv ready — just cd there and source .venv/bin/activate.
+
+You also need a ROM file. Ask the user for theirs. On this machine
+one exists at roms/pokemon_red.gb inside that directory.
+NEVER download or provide ROM files — always ask the user.
+
+### 2. Start the game server
+From inside the pokemon-agent directory with the venv activated, run
+pokemon-agent serve with --rom pointing to the ROM and --port 9876.
+Run it in the background with &.
+To resume from a saved game, add --load-state with the save name.
+Wait 4 seconds for startup, then verify with GET /health.
+
+### 3. Set up live dashboard for user to watch
+Use an SSH reverse tunnel via localhost.run so the user can view
+the dashboard in their browser. Connect with ssh, forwarding local
+port 9876 to remote port 80 on nokey@localhost.run. Redirect output
+to a log file, wait 10 seconds, then grep the log for the .lhr.life
+URL. Give the user the URL with /dashboard/ appended.
+The tunnel URL changes each time — give the user the new one if restarted.
+
+## Save and Load
+
+### When to save
+- Every 15-20 turns of gameplay
+- ALWAYS before gym battles, rival encounters, or risky fights
+- Before entering a new town or dungeon
+- Before any action you are unsure about
+
+### How to save
+POST /save with a descriptive name. Good examples:
+before_brock, route1_start, mt_moon_entrance, got_cut
+
+### How to load
+POST /load with the save name.
+
+### List available saves
+GET /saves returns all saved states.
+
+### Loading on server startup
+Use --load-state flag when starting the server to auto-load a save.
+This is faster than loading via the API after startup.
+
+## The Gameplay Loop
+
+### Step 1: OBSERVE — check state AND take a screenshot
+GET /state for position, HP, battle, dialog.
+GET /screenshot and save to /tmp/pokemon.png, then use vision_analyze.
+Always do BOTH — RAM state gives numbers, vision gives spatial awareness.
+
+### Step 2: ORIENT
+- Dialog/text on screen → advance it
+- In battle → fight or run
+- Party hurt → head to Pokemon Center
+- Near objective → navigate carefully
+
+### Step 3: DECIDE
+Priority: dialog > battle > heal > story objective > training > explore
+
+### Step 4: ACT — move 2-4 steps max, then re-check
+POST /action with a SHORT action list (2-4 actions, not 10-15).
+
+### Step 5: VERIFY — screenshot after every move sequence
+Take a screenshot and use vision_analyze to confirm you moved where
+intended. This is the MOST IMPORTANT step. Without vision you WILL get lost.
+
+### Step 6: RECORD progress to memory with PKM: prefix
+
+### Step 7: SAVE periodically
+
+## Action Reference
+- press_a — confirm, talk, select
+- press_b — cancel, close menu
+- press_start — open game menu
+- walk_up/down/left/right — move one tile
+- hold_b_N — hold B for N frames (use for speeding through text)
+- wait_60 — wait about 1 second (60 frames)
+- a_until_dialog_end — press A repeatedly until dialog clears
+
+## Critical Tips from Experience
+
+### USE VISION CONSTANTLY
+- Take a screenshot every 2-4 movement steps
+- The RAM state tells you position and HP but NOT what is around you
+- Ledges, fences, signs, building doors, NPCs — only visible via screenshot
+- Ask the vision model specific questions: "what is one tile north of me?"
+- When stuck, always screenshot before trying random directions
+
+### Warp Transitions Need Extra Wait Time
+When walking through a door or stairs, the screen fades to black during
+the map transition. You MUST wait for it to complete. Add 2-3 wait_60
+actions after any door/stair warp. Without waiting, the position reads
+as stale and you will think you are still in the old map.
+
+### Building Exit Trap
+When you exit a building, you appear directly IN FRONT of the door.
+If you walk north, you go right back inside. ALWAYS sidestep first
+by walking left or right 2 tiles, then proceed in your intended direction.
+
+### Dialog Handling
+Gen 1 text scrolls slowly letter-by-letter. To speed through dialog,
+hold B for 120 frames then press A. Repeat as needed. Holding B makes
+text display at max speed. Then press A to advance to the next line.
+The a_until_dialog_end action checks the RAM dialog flag, but this flag
+does not catch ALL text states. If dialog seems stuck, use the manual
+hold_b + press_a pattern instead and verify via screenshot.
+
+### Ledges Are One-Way
+Ledges (small cliff edges) can only be jumped DOWN (south), never climbed
+UP (north). If blocked by a ledge going north, you must go left or right
+to find the gap around it. Use vision to identify which direction the
+gap is. Ask the vision model explicitly.
+
+### Navigation Strategy
+- Move 2-4 steps at a time, then screenshot to check position
+- When entering a new area, screenshot immediately to orient
+- Ask the vision model "which direction to [destination]?"
+- If stuck for 3+ attempts, screenshot and re-evaluate completely
+- Do not spam 10-15 movements — you will overshoot or get stuck
+
+### Running from Wild Battles
+On the battle menu, RUN is bottom-right. To reach it from the default
+cursor position (FIGHT, top-left): press down then right to move cursor
+to RUN, then press A. Wrap with hold_b to speed through text/animations.
+
+### Battling (FIGHT)
+On the battle menu FIGHT is top-left (default cursor position).
+Press A to enter move selection, A again to use the first move.
+Then hold B to speed through attack animations and text.
+
+## Battle Strategy
+
+### Decision Tree
+1. Want to catch? → Weaken then throw Poke Ball
+2. Wild you don't need? → RUN
+3. Type advantage? → Use super-effective move
+4. No advantage? → Use strongest STAB move
+5. Low HP? → Switch or use Potion
+
+### Gen 1 Type Chart (key matchups)
+- Water beats Fire, Ground, Rock
+- Fire beats Grass, Bug, Ice
+- Grass beats Water, Ground, Rock
+- Electric beats Water, Flying
+- Ground beats Fire, Electric, Rock, Poison
+- Psychic beats Fighting, Poison (dominant in Gen 1!)
+
+### Gen 1 Quirks
+- Special stat = both offense AND defense for special moves
+- Psychic type is overpowered (Ghost moves bugged)
+- Critical hits based on Speed stat
+- Wrap/Bind prevent opponent from acting
+- Focus Energy bug: REDUCES crit rate instead of raising it
+
+## Memory Conventions
+| Prefix | Purpose | Example |
+|--------|---------|---------|
+| PKM:OBJECTIVE | Current goal | Get Parcel from Viridian Mart |
+| PKM:MAP | Navigation knowledge | Viridian: mart is northeast |
+| PKM:STRATEGY | Battle/team plans | Need Grass type before Misty |
+| PKM:PROGRESS | Milestone tracker | Beat rival, heading to Viridian |
+| PKM:STUCK | Stuck situations | Ledge at y=28 go right to bypass |
+| PKM:TEAM | Team notes | Squirtle Lv6, Tackle + Tail Whip |
+
+## Progression Milestones
+- Choose starter
+- Deliver Parcel from Viridian Mart, receive Pokedex
+- Boulder Badge — Brock (Rock) → use Water/Grass
+- Cascade Badge — Misty (Water) → use Grass/Electric
+- Thunder Badge — Lt. Surge (Electric) → use Ground
+- Rainbow Badge — Erika (Grass) → use Fire/Ice/Flying
+- Soul Badge — Koga (Poison) → use Ground/Psychic
+- Marsh Badge — Sabrina (Psychic) → hardest gym
+- Volcano Badge — Blaine (Fire) → use Water/Ground
+- Earth Badge — Giovanni (Ground) → use Water/Grass/Ice
+- Elite Four → Champion!
+
+## Stopping Play
+1. Save the game with a descriptive name via POST /save
+2. Update memory with PKM:PROGRESS
+3. Tell user: "Game saved as [name]! Say 'play pokemon' to resume."
+4. Kill the server and tunnel background processes
+
+## Pitfalls
+- NEVER download or provide ROM files
+- Do NOT send more than 4-5 actions without checking vision
+- Always sidestep after exiting buildings before going north
+- Always add wait_60 x2-3 after door/stair warps
+- Dialog detection via RAM is unreliable — verify with screenshots
+- Save BEFORE risky encounters
+- The tunnel URL changes each time you restart it
--- a/skills/leisure/find-nearby/SKILL.md
+++ b/skills/leisure/find-nearby/SKILL.md
@@ -0,0 +1,69 @@
+---
+name: find-nearby
+description: Find nearby places (restaurants, cafes, bars, pharmacies, etc.) using OpenStreetMap. Works with coordinates, addresses, cities, zip codes, or Telegram location pins. No API keys needed.
+version: 1.0.0
+metadata:
+  hermes:
+    tags: [location, maps, nearby, places, restaurants, local]
+    related_skills: []
+---
+
+# Find Nearby — Local Place Discovery
+
+Find restaurants, cafes, bars, pharmacies, and other places near any location. Uses OpenStreetMap (free, no API keys). Works with:
+
+- **Coordinates** from Telegram location pins (latitude/longitude in conversation)
+- **Addresses** ("near 123 Main St, Springfield")
+- **Cities** ("restaurants in downtown Austin")
+- **Zip codes** ("pharmacies near 90210")
+- **Landmarks** ("cafes near Times Square")
+
+## Quick Reference
+
+```bash
+# By coordinates (from Telegram location pin or user-provided)
+python3 SKILL_DIR/scripts/find_nearby.py --lat <LAT> --lon <LON> --type restaurant --radius 1500
+
+# By address, city, or landmark (auto-geocoded)
+python3 SKILL_DIR/scripts/find_nearby.py --near "Times Square, New York" --type cafe
+
+# Multiple place types
+python3 SKILL_DIR/scripts/find_nearby.py --near "downtown austin" --type restaurant --type bar --limit 10
+
+# JSON output
+python3 SKILL_DIR/scripts/find_nearby.py --near "90210" --type pharmacy --json
+```
+
+### Parameters
+
+| Flag | Description | Default |
+|------|-------------|---------|
+| `--lat`, `--lon` | Exact coordinates | — |
+| `--near` | Address, city, zip, or landmark (geocoded) | — |
+| `--type` | Place type (repeatable for multiple) | restaurant |
+| `--radius` | Search radius in meters | 1500 |
+| `--limit` | Max results | 15 |
+| `--json` | Machine-readable JSON output | off |
+
+### Common Place Types
+
+`restaurant`, `cafe`, `bar`, `pub`, `fast_food`, `pharmacy`, `hospital`, `bank`, `atm`, `fuel`, `parking`, `supermarket`, `convenience`, `hotel`
+
+## Workflow
+
+1. **Get the location.** Look for coordinates (`latitude: ... / longitude: ...`) from a Telegram pin, or ask the user for an address/city/zip.
+
+2. **Ask for preferences** (only if not already stated): place type, how far they're willing to go, any specifics (cuisine, "open now", etc.).
+
+3. **Run the script** with appropriate flags. Use `--json` if you need to process results programmatically.
+
+4. **Present results** with names, distances, and Google Maps links. If the user asked about hours or "open now," check the `hours` field in results — if missing or unclear, verify with `web_search`.
+
+5. **For directions**, use the `directions_url` from results, or construct: `https://www.google.com/maps/dir/?api=1&origin=<LAT>,<LON>&destination=<LAT>,<LON>`
+
+## Tips
+
+- If results are sparse, widen the radius (1500 → 3000m)
+- For "open now" requests: check the `hours` field in results, cross-reference with `web_search` for accuracy since OSM hours aren't always complete
+- Zip codes alone can be ambiguous globally — prompt the user for country/state if results look wrong
+- The script uses OpenStreetMap data which is community-maintained; coverage varies by region
--- a/skills/leisure/find-nearby/scripts/find_nearby.py
+++ b/skills/leisure/find-nearby/scripts/find_nearby.py
@@ -0,0 +1,184 @@
+#!/usr/bin/env python3
+"""Find nearby places using OpenStreetMap (Overpass + Nominatim). No API keys needed.
+
+Usage:
+    # By coordinates
+    python find_nearby.py --lat 36.17 --lon -115.14 --type restaurant --radius 1500
+
+    # By address/city/zip (auto-geocoded)
+    python find_nearby.py --near "Times Square, New York" --type cafe --radius 1000
+    python find_nearby.py --near "90210" --type pharmacy
+
+    # Multiple types
+    python find_nearby.py --lat 36.17 --lon -115.14 --type restaurant --type bar
+
+    # JSON output for programmatic use
+    python find_nearby.py --near "downtown las vegas" --type restaurant --json
+"""
+
+import argparse
+import json
+import math
+import sys
+import urllib.parse
+import urllib.request
+from typing import Any
+
+OVERPASS_URLS = [
+    "https://overpass-api.de/api/interpreter",
+    "https://overpass.kumi.systems/api/interpreter",
+]
+NOMINATIM_URL = "https://nominatim.openstreetmap.org/search"
+USER_AGENT = "HermesAgent/1.0 (find-nearby skill)"
+TIMEOUT = 15
+
+
+def _http_get(url: str) -> Any:
+    req = urllib.request.Request(url, headers={"User-Agent": USER_AGENT})
+    with urllib.request.urlopen(req, timeout=TIMEOUT) as r:
+        return json.loads(r.read())
+
+
+def _http_post(url: str, data: str) -> Any:
+    req = urllib.request.Request(
+        url, data=data.encode(), headers={"User-Agent": USER_AGENT}
+    )
+    with urllib.request.urlopen(req, timeout=TIMEOUT) as r:
+        return json.loads(r.read())
+
+
+def haversine(lat1: float, lon1: float, lat2: float, lon2: float) -> float:
+    """Distance in meters between two coordinates."""
+    R = 6_371_000
+    rlat1, rlat2 = math.radians(lat1), math.radians(lat2)
+    dlat = math.radians(lat2 - lat1)
+    dlon = math.radians(lon2 - lon1)
+    a = math.sin(dlat / 2) ** 2 + math.cos(rlat1) * math.cos(rlat2) * math.sin(dlon / 2) ** 2
+    return R * 2 * math.atan2(math.sqrt(a), math.sqrt(1 - a))
+
+
+def geocode(query: str) -> tuple[float, float]:
+    """Convert address/city/zip to coordinates via Nominatim."""
+    params = urllib.parse.urlencode({"q": query, "format": "json", "limit": 1})
+    results = _http_get(f"{NOMINATIM_URL}?{params}")
+    if not results:
+        print(f"Error: Could not geocode '{query}'. Try a more specific address.", file=sys.stderr)
+        sys.exit(1)
+    return float(results[0]["lat"]), float(results[0]["lon"])
+
+
+def find_nearby(lat: float, lon: float, types: list[str], radius: int = 1500, limit: int = 15) -> list[dict]:
+    """Query Overpass for nearby amenities."""
+    # Build Overpass QL query
+    type_filters = "".join(
+        f'nwr["amenity"="{t}"](around:{radius},{lat},{lon});' for t in types
+    )
+    query = f"[out:json][timeout:{TIMEOUT}];({type_filters});out center tags;"
+
+    # Try each Overpass server
+    data = None
+    for url in OVERPASS_URLS:
+        try:
+            data = _http_post(url, f"data={urllib.parse.quote(query)}")
+            break
+        except Exception:
+            continue
+
+    if not data:
+        return []
+
+    # Parse results
+    places = []
+    for el in data.get("elements", []):
+        tags = el.get("tags", {})
+        name = tags.get("name")
+        if not name:
+            continue
+
+        # Get coordinates (nodes have lat/lon directly, ways/relations use center)
+        plat = el.get("lat") or (el.get("center", {}) or {}).get("lat")
+        plon = el.get("lon") or (el.get("center", {}) or {}).get("lon")
+        if not plat or not plon:
+            continue
+
+        dist = haversine(lat, lon, plat, plon)
+
+        place = {
+            "name": name,
+            "type": tags.get("amenity", ""),
+            "distance_m": round(dist),
+            "lat": plat,
+            "lon": plon,
+            "maps_url": f"https://www.google.com/maps/search/?api=1&query={plat},{plon}",
+            "directions_url": f"https://www.google.com/maps/dir/?api=1&origin={lat},{lon}&destination={plat},{plon}",
+        }
+
+        # Add useful optional fields
+        if tags.get("cuisine"):
+            place["cuisine"] = tags["cuisine"]
+        if tags.get("opening_hours"):
+            place["hours"] = tags["opening_hours"]
+        if tags.get("phone"):
+            place["phone"] = tags["phone"]
+        if tags.get("website"):
+            place["website"] = tags["website"]
+        if tags.get("addr:street"):
+            addr_parts = [tags.get("addr:housenumber", ""), tags.get("addr:street", "")]
+            if tags.get("addr:city"):
+                addr_parts.append(tags["addr:city"])
+            place["address"] = " ".join(p for p in addr_parts if p)
+
+        places.append(place)
+
+    # Sort by distance, limit results
+    places.sort(key=lambda p: p["distance_m"])
+    return places[:limit]
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Find nearby places via OpenStreetMap")
+    parser.add_argument("--lat", type=float, help="Latitude")
+    parser.add_argument("--lon", type=float, help="Longitude")
+    parser.add_argument("--near", type=str, help="Address, city, or zip code (geocoded automatically)")
+    parser.add_argument("--type", action="append", dest="types", default=[], help="Place type (restaurant, cafe, bar, pharmacy, etc.)")
+    parser.add_argument("--radius", type=int, default=1500, help="Search radius in meters (default: 1500)")
+    parser.add_argument("--limit", type=int, default=15, help="Max results (default: 15)")
+    parser.add_argument("--json", action="store_true", dest="json_output", help="Output as JSON")
+    args = parser.parse_args()
+
+    # Resolve coordinates
+    if args.near:
+        lat, lon = geocode(args.near)
+    elif args.lat is not None and args.lon is not None:
+        lat, lon = args.lat, args.lon
+    else:
+        print("Error: Provide --lat/--lon or --near", file=sys.stderr)
+        sys.exit(1)
+
+    if not args.types:
+        args.types = ["restaurant"]
+
+    places = find_nearby(lat, lon, args.types, args.radius, args.limit)
+
+    if args.json_output:
+        print(json.dumps({"origin": {"lat": lat, "lon": lon}, "results": places, "count": len(places)}, indent=2))
+    else:
+        if not places:
+            print(f"No {'/'.join(args.types)} found within {args.radius}m")
+            return
+        print(f"Found {len(places)} places within {args.radius}m:\n")
+        for i, p in enumerate(places, 1):
+            dist_str = f"{p['distance_m']}m" if p["distance_m"] < 1000 else f"{p['distance_m']/1000:.1f}km"
+            print(f"  {i}. {p['name']} ({p['type']}) — {dist_str}")
+            if p.get("cuisine"):
+                print(f"     Cuisine: {p['cuisine']}")
+            if p.get("hours"):
+                print(f"     Hours: {p['hours']}")
+            if p.get("address"):
+                print(f"     Address: {p['address']}")
+            print(f"     Map: {p['maps_url']}")
+            print()
+
+
+if __name__ == "__main__":
+    main()
--- a/skills/mcp/native-mcp/SKILL.md
+++ b/skills/mcp/native-mcp/SKILL.md
@@ -321,6 +321,32 @@ mcp_servers:

 All tools from all servers are registered and available simultaneously. Each server's tools are prefixed with its name to avoid collisions.

+## Sampling (Server-Initiated LLM Requests)
+
+Hermes supports MCP's `sampling/createMessage` capability — MCP servers can request LLM completions through the agent during tool execution. This enables agent-in-the-loop workflows (data analysis, content generation, decision-making).
+
+Sampling is **enabled by default**. Configure per server:
+
+```yaml
+mcp_servers:
+  my_server:
+    command: "npx"
+    args: ["-y", "my-mcp-server"]
+    sampling:
+      enabled: true           # default: true
+      model: "gemini-3-flash" # model override (optional)
+      max_tokens_cap: 4096    # max tokens per request
+      timeout: 30             # LLM call timeout (seconds)
+      max_rpm: 10             # max requests per minute
+      allowed_models: []      # model whitelist (empty = all)
+      max_tool_rounds: 5      # tool loop limit (0 = disable)
+      log_level: "info"       # audit verbosity
+```
+
+Servers can also include `tools` in sampling requests for multi-turn tool-augmented workflows. The `max_tool_rounds` config prevents infinite tool loops. Per-server audit metrics (requests, errors, tokens, tool use count) are tracked via `get_mcp_status()`.
+
+Disable sampling for untrusted servers with `sampling: { enabled: false }`.
+
 ## Notes

 - MCP tools are called synchronously from the agent's perspective but run asynchronously on a dedicated background event loop
--- a/skills/media/DESCRIPTION.md
+++ b/skills/media/DESCRIPTION.md
@@ -1 +1,3 @@
-Media content extraction and transformation tools — YouTube transcripts, audio, video processing.
+---
+description: Skills for working with media content — YouTube transcripts, GIF search, music generation, and audio visualization.
+---
--- a/skills/media/gif-search/SKILL.md
+++ b/skills/media/gif-search/SKILL.md
--- a/skills/music-creation/heartmula/SKILL.md
+++ b/skills/music-creation/heartmula/SKILL.md
--- a/skills/music-creation/songsee/SKILL.md
+++ b/skills/music-creation/songsee/SKILL.md
--- a/skills/mlops/cloud/DESCRIPTION.md
+++ b/skills/mlops/cloud/DESCRIPTION.md
@@ -0,0 +1,3 @@
+---
+description: GPU cloud providers and serverless compute platforms for ML workloads.
+---
--- a/skills/mlops/cloud/lambda-labs/SKILL.md
+++ b/skills/mlops/cloud/lambda-labs/SKILL.md
--- a/skills/mlops/cloud/lambda-labs/references/advanced-usage.md
+++ b/skills/mlops/cloud/lambda-labs/references/advanced-usage.md
--- a/skills/mlops/cloud/lambda-labs/references/troubleshooting.md
+++ b/skills/mlops/cloud/lambda-labs/references/troubleshooting.md
--- a/skills/mlops/cloud/modal/SKILL.md
+++ b/skills/mlops/cloud/modal/SKILL.md
--- a/skills/mlops/cloud/modal/references/advanced-usage.md
+++ b/skills/mlops/cloud/modal/references/advanced-usage.md
--- a/skills/mlops/cloud/modal/references/troubleshooting.md
+++ b/skills/mlops/cloud/modal/references/troubleshooting.md
--- a/skills/mlops/evaluation/DESCRIPTION.md
+++ b/skills/mlops/evaluation/DESCRIPTION.md
@@ -0,0 +1,3 @@
+---
+description: Model evaluation benchmarks, experiment tracking, data curation, tokenizers, and interpretability tools.
+---
--- a/skills/mlops/evaluation/huggingface-tokenizers/SKILL.md
+++ b/skills/mlops/evaluation/huggingface-tokenizers/SKILL.md
--- a/Show More
+++ b/Show More