grill-me
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided plans and codebase content to generate feedback and update project documentation.
- Ingestion points: User plans, design descriptions, and codebase files are ingested during the interrogation session.
- Boundary markers: There are no explicit delimiters or specific instructions to treat user-provided plan text as data rather than instructions.
- Capability inventory: The skill has the capability to write and update markdown files, specifically
CONTEXT.md,CONTEXT-MAP.md, and ADRs in thedocs/adr/directory. - Sanitization: No explicit sanitization or validation of external content is described before it is interpolated into the interrogation or written to the documentation files.
- [COMMAND_EXECUTION]: The skill's documentation contains a reference to a local Node.js script (
node skills/skill-creator/scripts/validate-skill-library.mjs) used for manual validation of the skill library. This is a standard developer tool and does not indicate malicious runtime execution by the agent. - [SAFE]: No critical security issues were identified. The skill does not contain hardcoded credentials, persistence mechanisms, or obfuscated content designed to bypass safety filters.
Audit Metadata