mental-model-evaluator
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No executable code, scripts, or binary assets are included in the skill. The functionality is delivered entirely via natural language instructions.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided plans for analysis, which represents an indirect prompt injection surface. Ingestion points: User plans provided to the agent via trigger phrases in SKILL.md. Boundary markers: None; the skill does not use specific delimiters to separate user input from the analytical instructions. Capability inventory: None; the skill lacks any tools or code-based capabilities to perform actions like network requests or file modifications. Sanitization: None; the skill relies on the underlying model's safety guardrails, but the lack of executable capabilities makes this surface inert.
Audit Metadata