collaborative-engineering-agent
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of external repository content and instructions, creating an indirect prompt injection surface where untrusted data could influence agent behavior.
- Ingestion points: Local repository files, project-specific configuration, and task-specific instructions provided at runtime.
- Boundary markers: The skill lack strict boundary markers and explicitly states that "Repository instructions and the user's scope outrank this workflow," effectively allowing external content to override core instructions.
- Capability inventory: The agent is authorized to modify local files and perform external mutations (e.g., git operations) upon explicit user request.
- Sanitization: The skill mitigates risks through a structured "Delivery Loop" that mandates review and verification phases prior to completion.
Audit Metadata