learn

You can update AGENTS.md by analyzing the mistake, extracting a shareable lesson, identifying the correct documentation section, and appending an actionable row that codifies the anti-pattern and corrected behavior to prevent repeated errors.

Preventing repeated mistakes across sessions involves recording actionable lessons in AGENTS.md, ensuring the added entries are generalizable, correctly sectioned, and codify both the anti-pattern and the corrected behavior for future reference.

Yes, you can use this process for agent-training by extracting generalizable lessons from past mistakes and appending them as concise, actionable rows to the appropriate AGENTS.md section for governance and training purposes.

You ensure retrospective lessons are actionable and generalizable by extracting a concrete, shareable lesson from the mistake and formatting it as a concise documentation row that codifies the anti-pattern and the corrected behavior in the right context.

When updating AGENTS.md with process improvements, target appropriate sections such as General, TypeScript, Database, or Orchestration Model to ensure the lesson sits in the correct technical context.