Docs Cleanup

Docs Cleanup

Use this skill when the user asks for a documentation audit, cleanup pass, or stale-doc review.

Goals

• Find conflicting or outdated install/run instructions.
• Check whether examples still match current CLI/API behavior.
• Flag missing guidance that blocks onboarding, testing, or release tasks.
• Keep output focused on actionable findings with file references.

Workflow

1. Collect candidate docs:
   • README.md
   • docs/source/**/*.rst
   • CONTRIBUTING.md
   • .github/ templates/workflows when relevant to user instructions.
2. Validate install/run instructions against current repo behavior:
   • Build and test commands in docs vs commands in CMakeLists.txt, pyproject.toml, CI workflows, and scripts/.
   • Confirm paths, filenames, and prerequisites are still valid.
3. Validate examples:
   • Compare documented examples with samples/, test/, and current interfaces.
   • Check that command arguments and expected outputs are still plausible.
4. Identify gaps:
   • Missing prerequisites, platform caveats, troubleshooting notes, or release-process notes.
5. Report findings first, ordered by severity:
   • High: wrong instruction likely to fail.
   • Medium: confusing/outdated but recoverable.
   • Low: clarity/consistency improvements.

Output Format

• Findings list with severity, file, and exact issue.
• Proposed fix for each finding.
• Open questions or assumptions.
• If no findings: explicitly state that and note residual risk areas not fully validated.

Guardrails

• Prefer small documentation edits over broad rewrites.
• Avoid speculative claims; cite the source file/command that contradicts the docs.
• Do not edit generated docs under docs/_build/ or docs/doxygen/.