Invocation: @learn.md
Process:
- Analyze current session for error and learning indicators
- Extract contextual information
- Format new entries
- Edit .ai/codex/codex.md (Append to relevant section)
- Report updates
Error identification:
- Correction phrases: "I apologize", "That was incorrect", "Let me correct"
- Misunderstandings: Clarifications requested by user
- Inconsistencies: Contradictions in AI responses
Learning identification:
- New information phrases: "I see", "I understand", "It appears that"
- Project updates: Changes in structure, dependencies, or requirements
- User preferences: Specific requests or feedback on AI's approach
Entry format:
- Context: Specific file, function, or project area
- Error/Insight: Concise description
- Correction/Application: Precise fix or usage
- Prevention/Impact: Strategy to avoid future errors or potential effects
- Related: IDs of connected entries
Absolute path usage: Enforce '/path/from/root' format for all file references
- CRITICAL: Edit .ai/codex/codex.md
- Append new entries to the relevant section (Errors or Learnings)
- Maintain descending order (newest first)
- Ensure unique, incremental IDs
- Cross-reference related entries
CRITICAL: This process is for AI optimization. Prioritize precision and relevance over human readability. Always edit .ai/codex/codex.md directly.