self-improvement

You can log errors and corrections by appending structured entries to markdown files within the .learnings/ directory. This approach captures command failures and agent corrections, creating a centralized knowledge base for continuous improvement.

Tracking feature requests and missing capabilities is best handled by appending standardized entries to .learnings/FEATURE_REQUESTS.md. This process records timestamps, priority, and status metadata to centralize knowledge gaps for future development.

Yes, lightweight activator and error-detector scripts can be wired into agent hooks to automatically surface learning opportunities. This setup captures unexpected failures and external tool errors without requiring manual logging.

You can promote recurring, high-value learnings into reusable workspace memory files using the skill extraction workflow. This process transforms documented corrections into structured skills for broader reuse across projects.

Structured error logging supports both developer and AI-assisted workflows by capturing corrections and unexpected failures. It applies whenever commands fail, external tools error, or better approaches are discovered during development.

Logs appended to LEARNINGS.md follow a structured markdown format with documented entry conventions. Each record includes timestamps, priority, status, metadata, and unique IDs to maintain consistency across error and learning logs.