grill-me
Pass
Audited by Gen Agent Trust Hub on Jul 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill performs its stated purpose of document and plan review without any detected malicious patterns or unauthorized exfiltration.
- [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) to fetch issue data, which is a standard workflow for developer-oriented agents.
- [DATA_EXPOSURE]: The skill accesses local documents and GitHub issue content to perform plan reviews; this access is transparent and initiated by the user.
- [PROMPT_INJECTION]: The skill processes untrusted external data (GitHub issues and local files), creating a surface for indirect prompt injection. 1. Ingestion points: Data enters the context via reading local files and GitHub issue comments (SKILL.md). 2. Boundary markers: The instructions lack explicit delimiters or warnings to ignore embedded instructions in the ingested data. 3. Capability inventory: The skill can read local files and make GitHub API calls via the CLI. 4. Sanitization: No specific sanitization or validation logic is defined for the external document content. The sequential, interactive nature of the skill provides a natural human-in-the-loop check.
Audit Metadata