self-improvement

Error logging for continuous improvement captures command failures, API errors, and unexpected behavior within AI agent workflows. It records the error context and suggested fixes to prevent recurring mistakes and facilitate knowledge management.

Capturing user corrections and knowledge gaps is achieved through structured logging of discovered best practices and user feedback. This mechanism prevents recurring mistakes by recording the exact context of the failure for future reference.

Promoting insights to project memory elevates key learnings and corrections into persistent documentation like CLAUDE.md or AGENTS.md. This ensures that valuable knowledge is retained across sessions to prevent future occurrences of the same errors.

Yes, skill extraction facilitates the creation of new, reusable skills from valuable learnings. When an error recurs, the logged context and suggested fix can be promoted into a distinct skill to handle the scenario automatically.

Agent feedback integration supports proactive detection and logging of issues through agent hooks. This allows the system to automatically capture knowledge gaps and unexpected behavior as they occur during execution without manual intervention.