The Agent Skills Directory

[SAFE]: The skill follows best practices for development workflows. It uses standard developer tools (caliper) and provides clear, interactive prompts for user confirmation before writing or modifying files.
[COMMAND_EXECUTION]: The skill utilizes shell commands to execute the 'caliper' evaluation tool. These commands (caliper validate, caliper run, etc.) are appropriate for the skill's stated purpose of testing AI agent behavior.
[EXTERNAL_DOWNLOADS]: The skill mentions installing caliper-eval via pipx. This is a standard installation procedure for a well-known development tool and aligns with the skill's intended use case.
[DATA_EXPOSURE]: The evaluation examples in grill-skill.eval.yaml interact with temporary files in /tmp/, which is standard practice for testing and does not expose sensitive user data.

grill-skill