self-improvement

To log errors and learnings in an OpenClaw workspace, the skill captures command failures and user corrections into structured markdown files like LEARNINGS.md and ERRORS.md within the .learnings directory, enabling continuous AI improvement across sessions.

Continuous AI self-improvement via structured memory files is the process of recording corrections and missing features into markdown logs, then promoting broadly applicable learnings to CLAUDE.md and AGENTS.md for cross-session guidance.

You promote learnings to CLAUDE.md and AGENTS.md by identifying broadly applicable corrections from the .learnings directory and moving them into these workspace files, applying best practices across future OpenClaw sessions.

Yes, you can use workflow hooks for automated memory integration. The skill supports OpenClaw workflow automation via hooks, allowing it to automatically prompt logging after task completion without manual intervention.

Yes, the skill works with GitHub Copilot instructions. It promotes broadly applicable learnings and corrections to the .github/copilot-instructions.md file, ensuring cross-session guidance extends to your Copilot integration.

You should not use markdown files for logging learnings if your project requires high-frequency automated queries rather than human-readable cross-session guidance, as the structured markdown format prioritizes readability over database performance.