deep-grill
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a robust planning methodology designed to reduce requirement ambiguity through a two-phase interrogation workflow. This is a functional and benign use of an AI agent skill.\n- [SAFE]: File system operations are restricted to writing markdown decision records in a documented project directory (
docs/local/deliverables/deep-grill/). There is no evidence of unauthorized file access or modification of sensitive system files.\n- [PROMPT_INJECTION]: The skill processes user-supplied plans, which constitutes an ingestion point for untrusted data. While there are no specific sanitization steps, the turn-based 'one question at a time' constraint effectively prevents the agent from executing large-scale malicious instruction sets that might be hidden in a plan.\n- [DATA_EXFILTRATION]: No network activity or credential exposure was identified. The skill correctly identifies itself as not being for security auditing or credential management.
Audit Metadata