check-docs

Verify technical document claims against code, tests, and runtime behavior.

5|Updated Mar 18, 2026
One-click install
npx skills add https://github.com/sambeau/kanbanzai --skill check-docs-sambeau
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: check-docs
Source: https://github.com/sambeau/kanbanzai/tree/main/.kbz/skills/check-docs
Command: npx skills add https://github.com/sambeau/kanbanzai --skill check-docs-sambeau

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Documentation often contains unverified claims or hallucinations that diverge from the actual implementation. This skill identifies factual errors in technical documents by cross-checking with code, tests, and runtime behavior.

Core Features & Use Cases

  • Factual verification: checks each claim against the codebase, APIs, tests, and runtime behavior.
  • Code-example verification: runs or traces code examples to confirm outputs and API usage.
  • Reporting: emits a structured QA report with classifications (hallucination, unverified, stale, vague, inflated, promotional), and actionable remediation guidance.
  • Use Case: technical writers validating API docs against current library versions.

Quick Start

Run a document QA pass to identify factual claims and verify them against the codebase.

Frequently Asked Questions about check-docs

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

FAQPage Schema
How do I verify technical documentation accuracy against my codebase?

To verify technical documentation accuracy against a codebase, run a QA pass that maps each factual claim to a source of truth, executing or tracing code examples to confirm outputs and API usage.

How can I detect hallucinations and unverified claims in API references?

Detecting hallucinations in API references requires cross-checking document claims against the actual implementation, tests, and runtime behavior, producing a structured QA report to flag unverifiable or inconsistent details.

What's the best way to fact-check user guides for stale or vague information?

The best way to fact-check user guides for stale or vague information is to systematically validate each claim against the codebase, generating a QA report with classifications like stale, vague, or hallucination.

Can I validate design docs against runtime behavior and tests?

Yes, you can validate design docs against runtime behavior and tests by applying a verification process that requires code examples to be executable or traceable, ensuring claims match actual software behavior.

Does code-example verification work for tracing outputs in library documentation?

Code-example verification works for tracing outputs in library documentation by running or tracing the examples to confirm they produce the expected API usage and results without manual execution.