grill-autopilot
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Node.js scripts (e.g.,
grill-ledger.mjs) to manage the interview state and validate agent responses against a structured ledger. - [EXTERNAL_DOWNLOADS]: The setup instructions reference installing prerequisite skills from the author's repository and another well-known developer's repository on GitHub, which is standard practice for this platform.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied plans and Proxy AI answers. It utilizes a deterministic validator script to enforce specific field requirements and reversibility classifications, reducing the risk of unintended behavior from external data.
- [DATA_EXFILTRATION]: While the skill facilitates web research and multi-agent interaction, these actions are scoped to the primary design task. No evidence of harvesting sensitive local files, environment variables, or credentials was found.
Audit Metadata