What problem does it solve?
Documenting an unfamiliar codebase is slow and error-prone, and AI-generated docs often invent deployment steps or configuration details that do not exist. This Skill produces documentation where every operational claim cites its source file and carries a confidence label, so gaps are recorded instead of hallucinated.
Core Features & Use Cases
- Evidence-grounded documentation: Generates overview, operation, deployment, learning, architecture, API/data, troubleshooting, and maintenance manuals where each claim cites a source file with a Verified / Inferred / Unknown / Needs confirmation label.
- Machine-readable index.json: Produces an
index.json manifest plus an _evidence/ map (source map, assumptions, generation log) so AI agents can consume the documentation package.
- Quality checks: Verifies that referenced paths exist, commands are sourced, no secret values leak, and undocumented gaps are explicitly noted.
- Use Case: Point it at a newly inherited repository to produce a
study-docs/ onboarding package whose deployment instructions are traceable to actual config files, with missing pieces flagged rather than fabricated.
Quick Start
Use tracedocs to generate evidence-grounded study docs for this repository and write the output to study-docs/.
