feynman-auditor
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses persona-defining rules (e.g., 'RULE 0: QUESTION EVERYTHING') and an 'Anti-Hallucination Protocol' to govern the auditor's reasoning process. These instructions are task-specific and do not attempt to bypass general AI safety guardrails or extract system prompts.
- [COMMAND_EXECUTION]: The execution process includes a 'Verification Gate' (Phase 5) which directs the agent to utilize native language tools (e.g.,
forge test,cargo test,go test) to validate audit hypotheses. This functionality is intended for legitimate code verification purposes and is a core component of the skill's utility. - [DATA_EXPOSURE]: The skill defines local file paths for reporting results (e.g.,
.audit/findings/feynman-verified.md). This is a standard reporting practice and does not involve unauthorized access to sensitive system directories or credentials.
Audit Metadata