feat: improve memory prioritization — user preferences over procedural knowledge

Inspired by OpenAI Codex's memory prompt improvements (openai/codex#14493)
which focus memory writes on user preferences and recurring patterns
rather than procedural task details.

Key insight: 'Optimize for reducing future user steering — the most
valuable memory prevents the user from having to repeat themselves.'

Changes:
- MEMORY_GUIDANCE (prompt_builder.py): added prioritization hierarchy
  and the core principle about reducing user steering
- MEMORY_SCHEMA (memory_tool.py): reordered WHEN TO SAVE list to put
  corrections first, added explicit PRIORITY guidance
- Memory nudge (run_agent.py): now asks specifically about preferences,
  corrections, and workflow patterns instead of generic 'anything'
- Memory flush (run_agent.py): now instructs to prioritize user
  preferences and corrections over task-specific details

agent/prompt_builder.py

@@ -73,9 +73,15 @@ DEFAULT_AGENT_IDENTITY = (
 MEMORY_GUIDANCE = (
     "You have persistent memory across sessions. Save durable facts using the memory "
     "tool: user preferences, environment details, tool quirks, and stable conventions. "
-    "Memory is injected into every turn, so keep it compact. Do NOT save task progress, "
-    "session outcomes, or completed-work logs to memory; use session_search to recall "
-    "those from past transcripts."
+    "Memory is injected into every turn, so keep it compact and focused on facts that "
+    "will still matter later.\n"
+    "Prioritize what reduces future user steering — the most valuable memory is one "
+    "that prevents the user from having to correct or remind you again. "
+    "User preferences and recurring corrections matter more than procedural task details.\n"
+    "Do NOT save task progress, session outcomes, completed-work logs, or temporary TODO "
+    "state to memory; use session_search to recall those from past transcripts. "
+    "If you've discovered a new way to do something, solved a problem that could be "
+    "necessary later, save it as a skill with the skill tool."
 )
 
 SESSION_SEARCH_GUIDANCE = (

run_agent.py

@@ -3542,7 +3542,8 @@ class AIAgent:
 
         flush_content = (
             "[System: The session is being compressed. "
-            "Please save anything worth remembering to your memories.]"
+            "Save anything worth remembering — prioritize user preferences, "
+            "corrections, and recurring patterns over task-specific details.]"
         )
         _sentinel = f"__flush_{id(self)}_{time.monotonic()}"
         flush_msg = {"role": "user", "content": flush_content, "_flush_sentinel": _sentinel}
@@ -4541,8 +4542,9 @@ class AIAgent:
             self._turns_since_memory += 1
             if self._turns_since_memory >= self._memory_nudge_interval:
                 user_message += (
-                    "\n\n[System: You've had several exchanges in this session. "
-                    "Consider whether there's anything worth saving to your memories.]"
+                    "\n\n[System: You've had several exchanges. Consider: "
+                    "has the user shared preferences, corrected you, or revealed "
+                    "something about their workflow worth remembering for future sessions?]"
                 )
                 self._turns_since_memory = 0
 

tools/memory_tool.py

@@ -439,11 +439,13 @@ MEMORY_SCHEMA = {
         "Memory is injected into future turns, so keep it compact and focused on facts "
         "that will still matter later.\n\n"
         "WHEN TO SAVE (do this proactively, don't wait to be asked):\n"
+        "- User corrects you or says 'remember this' / 'don't do that again'\n"
         "- User shares a preference, habit, or personal detail (name, role, timezone, coding style)\n"
         "- You discover something about the environment (OS, installed tools, project structure)\n"
-        "- User corrects you or says 'remember this' / 'don't do that again'\n"
         "- You learn a convention, API quirk, or workflow specific to this user's setup\n"
         "- You identify a stable fact that will be useful again in future sessions\n\n"
+        "PRIORITY: User preferences and corrections > environment facts > procedural knowledge. "
+        "The most valuable memory prevents the user from having to repeat themselves.\n\n"
         "Do NOT save task progress, session outcomes, completed-work logs, or temporary TODO "
         "state to memory; use session_search to recall those from past transcripts.\n"
         "If you've discovered a new way to do something, solved a problem that could be "