driangle/ai-powers/audit-docs

Claude Code and agent skill by driangle/ai-powers

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Audit documentation coverage for a project: discover CLI commands, public APIs, configuration options, and features, then cross-reference against documentation to find gaps, stale content, and missing sections. Use when the user wants to check if docs are up to date, find undocumented features, or verify doc completeness.

Install

npx vskill@latest install driangle/ai-powers/audit-docs

SKILL.md

Audit Documentation

Discover all documented surfaces of the project (CLI commands, public APIs, configuration, features), then cross-reference against the actual documentation to identify gaps, outdated content, and missing sections.

Instructions

Arguments in $ARGUMENTS are optional flags:

  • --fix — after reporting gaps, update the documentation files to fill them
  • --scope <area> — limit the audit to a specific area (e.g., cli, api, config, features)
  • --verbose — include per-item detail in the report
  • --since-last-release — only check items added since the last git tag

If no arguments are provided, run a full audit and produce a report (no fixes).

Phase 1: Understand the project

Before auditing, understand what kind of project this is and what documentation surfaces exist.

  1. Read project metadata: Check package.json, Cargo.toml, pyproject.toml, go.mod, Makefile, or similar to understand the project type, name, and structure.
  2. Find documentation: Look for docs in common locations:
  • docs/, doc/, documentation/
  • README.md, CHANGELOG.md
  • Doc site sources (e.g., apps/docs/, website/, site/, docusaurus/, .vitepress/, mkdocs.yml)
  • API docs (e.g., api-docs/, generated docs)
  • Man pages or help text
  1. Identify documentable surfaces — determine which of these apply to the project:
  • CLI commands & flags (if the project has a CLI)
  • Public API (exported functions, classes, types for libraries)
  • Configuration options (config files, environment variables)
  • Web features / UI pages (if the project has a web interface)
  • REST/GraphQL/RPC endpoints (if the project exposes an API)
Phase 2: Discover actual surfaces

For each documentable surface identified in Phase 1, discover what actually exists in the code.

CLI commands (if applicable)
  1. Build the project if needed to get a working binary.
  2. Run <binary> --help and parse available commands.
  3. For each command, run <binary> <command> --help to capture usage, description, and all flags.
  4. Recurse into subcommands.
  5. Capture global flags.
Public API (if applicable)
  1. Find all exported symbols: functions, classes, types, constants.
  2. Check for JSDoc/docstrings/doc comments on exports.
  3. Note which exports lack documentation.
Configuration (if applicable)
  1. Find config file schemas, types, or parsing code.
  2. Extract all supported config keys, their types, and defaults.
  3. Check for environment variable support.
Web features / endpoints (if applicable)
  1. Find route definitions and page components.
  2. Find API endpoint handlers.
  3. Compile a list of pages, routes, and endpoints.
Phase 3: Cross-reference against documentation

For each discovered item, check whether it is documented:

  1. Is it mentioned at all? Search the documentation for references to each item.
  2. Is the documentation accurate? Compare documented behavior (flag names, defaults, types, descriptions) against actual code.
  3. Are there examples? Check if usage examples exist for key features.
  4. Are there orphaned docs? Look for documentation that references items that no longer exist in the code.

Also check documentation infrastructure:

  • Navigation/sidebar: Do all listed pages exist? Are there orphan pages not linked from navigation?
  • Internal links: Are cross-references between doc pages valid?
  • Version references: Do version numbers in docs match the actual project version?
Phase 4: Generate report

Produce a structured markdown report:

## Documentation Audit Report
### Summary
For each surface area, report:
- Total items: X (documented: Y, missing: Z, stale: W)
### Undocumented Items
For each undocumented item:
- Item name and type
- Where it's defined in code
- Which doc file(s) should cover it
### Stale / Inaccurate Documentation
Items where docs don't match reality:
- What the docs say vs what the code does
- File and location of the stale content
### Documentation Infrastructure Issues
- Broken links or references
- Orphan pages (exist but not linked)
- Missing pages (linked but don't exist)
- Version mismatches
### Top Recommendations
Prioritized list of the most impactful documentation improvements

Print the report to stdout.

Phase 5: Fix gaps (only if --fix was passed)

If $ARGUMENTS contains --fix:

  1. For each gap identified in Phase 4, update the relevant documentation file.
  2. Follow the existing documentation style and formatting conventions in each file.
  3. For new sections, match the structure and tone of surrounding content.
  4. After making changes, list all modified files.

If --fix was NOT passed, end with: "Run /audit-docs --fix to automatically update documentation."

Error Handling
  • If a build step fails, report the error and continue with source-level analysis.
  • If a command's --help output can't be parsed, note it in the report and continue.
  • If a documentation file doesn't exist, note it as a gap.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/driangle/ai-powers/audit-docs/badge)](https://verified-skill.com/skills/driangle/ai-powers/audit-docs)

Something wrong with this skill or its page? Tell us.