Review Doc Consistency
Ensures documents align with established guidelines, flagging inconsistencies in style, tone, and factual accuracy.
Install on your platform
We auto-selected Claude Code based on this skill’s supported platforms.
Run in terminal (recommended)
claude mcp add review-doc-consistency npx -- -y @trustedskills/review-doc-consistency
Or manually add to ~/.claude/settings.json
{
"mcpServers": {
"review-doc-consistency": {
"command": "npx",
"args": [
"-y",
"@trustedskills/review-doc-consistency"
]
}
}
}Requires Claude Code (claude CLI). Run claude --version to verify your install.
About This Skill
What it does
This skill, Review Doc Consistency, systematically identifies inconsistencies between documentation and code within a project's README and documentation files (located in docs/**/*.md). It prioritizes discrepancies where documentation conflicts with source code, configuration files, or contract definitions (like OpenAPI/proto/schema/TS types), treating these as the single source of truth. The skill aims to output at least 30 identified issues, categorized by severity, and provides evidence for each finding.
When to use it
- Maintaining accurate API documentation: Ensure your API documentation accurately reflects the current code implementation.
- Validating development guides: Confirm that instructions in development guides are consistent with how the code actually works.
- Improving security posture: Identify and address any inconsistencies related to security configurations documented versus implemented.
- Standardizing project documentation: Enforce consistency across all documents, including READMEs and API references.
Key capabilities
- Document Scanning: Scans
README.md(root directory) and files within thedocs/**/*.mddirectories, as well as contract files (OpenAPI/proto/GraphQL schema/TS types). - Code-Documentation Comparison: Compares documentation claims with corresponding code implementations to identify mismatches in behavior, default values, or missing information.
- Severity Level Assignment: Categorizes issues based on severity levels: P0 (Security/Serious Misleading), P1 (Core Functionality Inconsistency), P2 (Incomplete Examples/Naming), P3 (Minor Wording Issues), and Pending Evidence.
- Evidence-Based Reporting: Provides specific code or configuration file locations as evidence for each identified inconsistency.
- Output Formatting: Generates a
doc-consistency.mdreport file and a structureddoc-consistency.jsonissue list.
Example prompts
- "Review the documentation for inconsistencies with the codebase."
- "Check all API documentation against the OpenAPI schema."
- "Find any discrepancies between the README and the current code implementation."
Tips & gotchas
- Contracts are authoritative: Remember that contract files (OpenAPI, proto, etc.) take precedence over documentation.
- Prioritization: Security-related inconsistencies (P0 severity) will be prioritized.
- Parallel Execution: The skill is designed to be run in parallel across different document types or modules for faster review times.
Tags
TrustedSkills Verification
Unlike other registries that point to live repositories, TrustedSkills pins every skill to a verified commit hash. This protects you from malicious updates — what you install today is exactly what was reviewed and verified.
Security Audits
| Gen Agent Trust Hub | Pass |
| Socket | Pass |
| Snyk | Pass |
🌐 Community
Passed automated security scans.