The Agent Skills Directory

[SAFE]: Documentation verification framework. The skill implements a robust procedure for auditing documentation by cross-referencing symbols, signatures, CLI flags, and behavioral claims with the actual source code. This is a defensive measure designed to eliminate technical inaccuracies and AI hallucinations.
[SAFE]: Credential and secret management. The references/code-samples.md file provides clear instructions to avoid hardcoded credentials in documentation, mandating the use of safe placeholders like RFC 5737 reserved IPs and standard dummy keys (e.g., YOUR_API_KEY). This promotes secure documentation practices.
[SAFE]: Trusted external research and guides. The skill references well-known research institutions and documentation standards (including arXiv, ACM, USENIX, Google, and the RFC Editor) for technical guidance. These links are informative and present no security risk.
[SAFE]: Verification procedure grounding. The skill's core logic relies on reading local implementation details and argument parser sources to confirm documentation accuracy. This ensures the agent's output is grounded in the project context rather than relying on potentially outdated or incorrect training data.

docs-guard