The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install a Python package directly from a public GitHub repository (github.com/joaquinhuigomez/agent-eval) using a specific commit hash. This introduces a dependency on code from a source that is not recognized as a trusted organization or well-known service.
[COMMAND_EXECUTION]: The core functionality described involves using the Bash tool to run a CLI that executes arbitrary shell commands (e.g., pytest, npm run build) specified in user-provided YAML task definitions.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data that could influence agent behavior:
Ingestion points: YAML task definitions located in the tasks/ directory.
Boundary markers: None mentioned; there are no instructions to the agent to ignore embedded commands or logic within the YAML files.
Capability inventory: The skill uses the Bash, Write, and Edit tools to modify the environment and execute commands based on the YAML input.
Sanitization: There is no evidence of validation or sanitization for the commands or patterns defined in the judge section of the task files.

agent-eval