good-spec
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Data Exposure & Exfiltration] (SAFE): The skill accesses project configuration files (e.g., package.json) to understand context but explicitly prohibits the agent from handling or committing secrets and environmental files.
- [Unverifiable Dependencies & Remote Code Execution] (SAFE): No remote scripts or external package installations are initiated by this skill.
- [Indirect Prompt Injection] (LOW): The skill processes untrusted project data and user input to generate documentation. Evidence Chain: 1. Ingestion points: package.json, SPEC.md, and user prompts. 2. Boundary markers: Implements a 3-stage boundary system (Always/Ask/Never) to restrict agent behavior. 3. Capability inventory: Local file system read and write operations. 4. Sanitization: Includes a validation phase with a checklist and AI 'Red Flags' to verify generated content.
Audit Metadata