learn

To extract lessons learned from conversations, analyze the chat history and append applicable learnings to assistant configuration files like CLAUDE.md or GEMINI.md. Each learning is validated for relevance and compatibility, then routed by topic and size to the appropriate config.

Yes, you can persist conversation learnings to multiple AI assistant configs including Claude Code, Gemini, AGENTS.md, Cursor, Copilot, Windsurf, and Continue configurations. Learnings are routed to the relevant configuration files based on the specific topic and context.

The best way to organize extracted learnings as content grows is by refactoring them into skills. The system routes learnings by topic, size, and relevance, offering refactoring into skills when necessary to maintain configuration file efficiency and clarity.

Routing learnings by topic and relevance works by validating each extracted learning for compatibility and relevance, then appending or organizing it into the correct AI assistant configuration file. This ensures learnings are persisted to the most applicable config.

No, manual validation is not needed before appending learnings to AGENTS.md. The system automatically validates each learning for relevance and compatibility, then appends or organizes it into the configuration based on routing rules.

Persisting conversation learnings into AI configuration files ensures that extracted lessons are permanently integrated into assistant configs. This allows the AI assistant to continuously apply validated learnings in future interactions, rather than losing them when chat history is cleared.