docs-navigator

Navigating project documentation efficiently involves performing a QMD search when available, or falling back to a local docs/index.md file, to read only the 1-3 most relevant documents and minimize context usage in unfamiliar codebases.

Searching documentation in an unfamiliar codebase is best handled by performing a QMD search to surface ranked results, or falling back to a local docs/index.md file to match keywords and identify the most pertinent documents for your task.

To find relevant codebase docs when QMD search is unavailable, consult the local docs/index.md file and match keywords to identify the most pertinent documents, then read 1-3 of them fully to ground your decisions.

This documentation search workflow requires no external dependencies; it enforces a strict fallback mechanism that uses a local docs/index.md file when QMD search is unavailable, requiring no additional setup.

Reading only 1-3 documents before acting minimizes context bloat in AI tasks, enabling faster understanding of unfamiliar codebases and conventions while grounding decisions in the most relevant materials without overwhelming the model.