Where this fits

You're two days from a release. The CLI has three new flags, but the README still shows the old ones. A user opens the docs, hits a missing command, and files a support ticket. This is the classic docs-drift problem. The gza-docs-review skill gives your agent a structured process to catch these gaps before users do. It reads your documentation, verifies it against actual implementation output (like --help), and flags missing or inaccurate information.

Why agents benefit

• Executes real verification: The skill runs Bash(uv run --help) to compare CLI help text against what's written, so the agent doesn't just skim prose—it checks facts.
• Maps the doc tree: Uses ls, Glob, and Grep to discover every documentation file, ensuring nothing is overlooked (including CHANGELOGs and CONTRIBUTING guides).
• Structured process: Follows a step-by-step method that mirrors a human reviewer's workflow—discover, read, verify—so results are consistent and reproducible.
• Time-aware checks: Includes a timestamp command to log when the review was performed, useful for tracking doc freshness in CI or audits.
• Zero hallucination risk: Since the skill forces the agent to look at actual files and command output, it won't invent missing features or misrepresent behavior.

Practical scenarios

Pre-release documentation audit

Before shipping a new version, run the skill against your project. The agent reads the CHANGELOG and compares the documented changes against the actual --help output. It will report any version bumps mentioned in docs but missing from code, or new flags not yet documented.

Onboarding guide gap detection

Your README claims a five-minute quick start, but the actual steps require an extra config. The agent reads the quick start, tries the commands via --help, and identifies places where the tutorial silently fails or misses required steps.

Third-party tool evaluation

Evaluating a new library? Use this skill to check if its docs are trustworthy. The agent will map the docs and cross-check them against the tool's own help output, giving you a confident go/no-go signal before integrating.

Add it to your agent workflow

Install the skill with:

npx skills add mhawthorne/gza --skill gza-docs-review

Then ask your agent to "review the documentation for accuracy and missing features." The agent will automatically invoke the skill, run the discovery steps, and produce a structured report with specific file and line references for every issue found. You can also point it at a subfolder or specific doc set to focus the review.

Compare before adopting

Before adopting this skill, look at its quality signals—the GitHub repo has 11 stars and is maintained by a single author. Check the last commit date to gauge freshness. Also compare with alternatives like generic documentation linters or manual code-review plugins. The key differentiator here is the use of --help verification and the tool's explicit process for finding missing info, not just typos. Evaluate whether the skill integrates with your existing agent harness (it's designed for skill-compatible agents like Claude or open-source frameworks).

Why it is worth tracking

This skill is a good example of what a focused automation skill should be—small, deterministic, and useful. Its star count is modest, but the process it encodes is valuable and likely to evolve as more developers contribute edge cases. Watch for updates that expand beyond CLI tools to cover REST APIs or config schemas. If you maintain open-source projects or work in a team that sees frequent docs drift, this is worth evaluating now. You can also explore similar skills on the OpenAgentSkill hub to see how it compares with other documentation-focused agents.