claudeception

Claudeception automatically extracts reusable knowledge from non-obvious debugging, trial-and-error discoveries, and workflow optimizations. It evaluates these work sessions and codifies the hard-won knowledge into Claude Code skills for future reuse.

You can automate skill extraction using Claudeception's activation hooks, which evaluate work sessions at startup and after completion. You can also manually trigger extraction by typing /claudeception or 'save this as a skill' to generate a new Claude Code skill.

Extracted workflow knowledge is stored using standard Claude Code skill templates. The SKILL.md format requires frontmatter containing name, description, author, and version, alongside a minimal Markdown body and optional scripts to govern activation and references.

Yes, Claudeception supports extracting skills after optimizing tooling and configuration tasks. It evaluates workflow optimizations across coding, config, and tooling tasks, generating a new skill if the discovered knowledge is deemed valuable for future reuse.

No, manual triggering is not required for every debugging session. Claudeception supports automatic evaluation hooks at startup and after sessions, allowing it to autonomously detect and extract valuable debugging knowledge without requiring a manual /claudeception command.

A limitation of automating continuous learning is that extraction requires non-obvious debugging or meaningful workflow optimizations to trigger. Routine coding tasks will not generate skills, and the system relies on standard SKILL.md frontmatter to properly govern activation and storage.