knowledge

To check your knowledge base for stale and unreferenced artifacts, you can lint your raw/ and compiled/ directories. This process scans the knowledge corpus and produces a read-only, actionable report detailing issues like missing types and oversized files without modifying anything.

You can lint a raw knowledge directory by executing the shared knowledge-lint script via node scripts/knowledge-lint.js .claude-code-hermit from the plugin install directory. This generates a concise, actionable remediation plan for any issues found, operating in a read-only mode to prevent file modifications.

Yes, you can audit knowledge directories without modifying files. The linting script runs in a read-only mode that reports findings across raw/ and compiled/ folders, providing actionable advice for issues like oversized artifacts and missing types while ensuring no data is altered.

The best way to identify oversized and missing-type artifacts is to run a dedicated knowledge-lint script across the corpus. It evaluates the directories and outputs a concise, actionable remediation plan to help you clean up the knowledge base effectively.

No, the knowledge lint script does not fix stale artifacts automatically. It operates as a read-only audit tool that identifies unreferenced, missing-type, and oversized artifacts, producing a report with actionable advice so you can manually apply the necessary remediation steps.