spec-drift
Pass
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security risks were detected. The skill documentation and instructions emphasize a 'Confirmation Protocol' that prevents the agent from making destructive or significant changes to the specification files without explicit user approval (SKILL.md).
- [PROMPT_INJECTION]: The skill processes untrusted data such as existing source code and external requirements to generate or update specifications, which represents an indirect prompt injection surface.
- Ingestion points: Existing codebase (via audit/create) and requirements text/tickets (via create) in SKILL.md.
- Boundary markers: Not explicitly mentioned in the instructions.
- Capability inventory: The agent has access to Write, Edit, and Bash tools for file management and analysis.
- Sanitization: No specific sanitization logic is described; however, the risk is mitigated by the mandatory confirmation protocol which requires the agent to show diffs and wait for user approval before applying any generated content.
Audit Metadata