project-spec
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill establishes a rigid framework for AI agent interactions by defining 'Critical Rules' that prioritize explicit project documentation over assumptions. This reduces the risk of unintended agent behavior or 'hallucinations' during project execution.
- [SAFE]: No evidence of data exfiltration, remote code execution, or credential theft was found. The skill does not perform network operations or access sensitive system paths.
- [SAFE]: The skill incorporates security-conscious requirements into its 'Definition of Done' templates, explicitly requiring the agent to verify that no credentials or secrets are present in the final code.
- [SAFE]: The 'Authority' hierarchy defined in the skill (CLAUDE.md > skills > chat) is a safety mechanism designed to ensure the agent adheres to the verified project specification above all other inputs.
Audit Metadata