grill-skill
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the 'caliper-eval' package via pipx. This is a standard dependency for the skill's primary function of running agent evaluations.
- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute 'caliper' CLI commands (validate, run, list, report). These commands are necessary for the skill's core purpose of measuring and iterating on skill performance.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external SKILL.md files to generate evaluation tasks. This represents a data ingestion point for untrusted content. The skill mitigates this risk by summarizing the content for the user and requiring explicit confirmation before generating any test specifications.
- Ingestion points: User-provided path to a SKILL.md file or automatic detection in the current directory.
- Boundary markers: The agent is instructed to summarize its understanding and wait for user confirmation before proceeding to Phase 2.
- Capability inventory: The agent has access to Bash, Read, Write, and Edit tools to create and execute evaluation specs.
- Sanitization: The skill relies on human-in-the-loop confirmation during the interview phase to ensure the generated test cases are accurate and safe.
Audit Metadata