self-improvement

To log errors and capture learnings for AI agents, record insights, errors, and corrections in a .learnings/ directory using YAML and Markdown formats for structured metadata management. This enables continuous improvement by systematically tracking agent experiences.

Continuous learning capture for AI coding agents is the process of recording errors, insights, and feature requests in structured YAML and Markdown logs to facilitate ongoing skill development and self-improvement across development workflows.

Yes, continuous improvement tracking works across multiple AI coding agents including Claude Code, Codex, Copilot, and OpenClaw, allowing learnings and error logs captured in the .learnings/ directory to be shared and applied broadly.

To convert captured learnings into reusable skills, use the skill extraction feature to transform valuable insights logged in YAML and Markdown formats into structured skill definitions that can be promoted to project memory for broader applicability.

AI agent error logging and feature tracking require YAML and Markdown file formats for structured logging and skill extraction. These formats support metadata management within the .learnings/ directory to organize learnings, errors, and feature requests.

Yes, you can track feature requests and promote useful learnings to project memory. The system tracks user requests for new features and elevates valuable insights to project memory for broader applicability across AI agent workflows.