quality-validation
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION] (LOW): Authoritative Behavioral Control. The skill uses coercive language such as 'Iron Law' and 'Non-negotiable' to override agent heuristics regarding task completion. This style of framing mirrors techniques used to bypass standard safety constraints.
- [PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface. The 'Gate Function' requires the agent to read and process full output from external commands. 1. Ingestion points: Full output from verification commands (SKILL.md). 2. Boundary markers: Absent; there are no instructions to use delimiters or treat tool output as untrusted. 3. Capability inventory: The skill directs the execution of shell commands (tests, builds, linters). 4. Sanitization: Absent; the agent is instructed to read and rely on the raw output directly.
Audit Metadata