docs-guard

Verify documentation claims, code samples, and API references against source code.

Updated Jun 22, 2026
One-click install
npx skills add https://github.com/aashahin/ai-skills --skill docs-guard-aashahin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: docs-guard
Source: https://github.com/aashahin/ai-skills/tree/main/docs-guard
Command: npx skills add https://github.com/aashahin/ai-skills --skill docs-guard-aashahin

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? AI-generated documentation often contains hallucinated functions, wrong signatures, broken code samples, and unverifiable claims that readers cannot distinguish from accurate docs. This Skill audits READMEs, API references, docstrings, changelogs, and tutorials against the actual source code before they ship. ## Core Features & Use Cases - Claim Verification: Extracts every referenced symbol, flag, endpoint, config key, and version from docs and checks each against the definition site, CLI parser, route table, or changelog. - Code Sample Auditing: Validates that samples have resolvable imports, real API signatures, no hardcoded local paths or credentials, and cover error paths. - Drift Detection: Greps all doc surfaces for renamed or removed symbols after code changes, and flags paraphrased docstrings, filler prose, and broken internal links. - Use Case: After an agent writes a README for a new API, run a review pass to catch a documented flag that does not exist in the argument parser and a code sample calling a function with the wrong argument order. ## Quick Start Review the README and docstrings I just generated and verify every function, flag, and code sample against the actual source code.

Frequently Asked Questions about docs-guard

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check if AI-generated documentation is accurate?▼

Extract every claim from the docs — functions, flags, endpoints, config keys, versions — and verify each against its source of truth: the definition site, argument parser, route table, or changelog. This Skill provides a structured verification procedure and review checklist for exactly that pass.

How to review code samples in documentation before publishing?▼

Check that imports resolve, API signatures match the real definitions, samples run on a clean machine without hardcoded paths or credentials, and shown output matches actual output. The Skill's code-samples reference defines a full shippability checklist including error-path coverage.

Can this review docstrings, PHPDoc, and JSDoc comments?▼

Yes, it audits in-code documentation for paraphrased signatures, @param name and type drift, incorrect @throws lists, and wrong @since versions. It treats wrong docstrings that feed generated reference sites with the same severity as wrong READMEs.

When should I not use documentation review guardrails?▼

Do not use it for production code review, test review, marketing copy, prose style editing, or documentation site theming. It verifies what docs claim about code, not the code itself, and does not enforce tone or style guides.

Why does AI-generated documentation contain wrong API references?▼

Models document from memory of how APIs usually look rather than reading the code in front of them, producing valid-looking but hallucinated invocations especially for infrequent APIs. Verification against the actual source is the only reliable detection method.