The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run a CLI tool (agent-eval) which in turn executes shell commands defined in the judge section of YAML task files (e.g., pytest, npm run build).
[EXTERNAL_DOWNLOADS]: The skill references an external repository (github.com/joaquinhuigomez/agent-eval) for the installation of the evaluation tool. This repository is not from a known trusted organization.
[REMOTE_CODE_EXECUTION]: By executing the agent-eval tool and processing external YAML task definitions, there is a risk of remote code execution if the task files or the tool itself are malicious.
[PROMPT_INJECTION]: The skill processes untrusted YAML task definitions which include prompt and judge commands, creating an indirect prompt injection surface. * Ingestion points: YAML files in the tasks/ directory. * Boundary markers: None identified in the provided instructions. * Capability inventory: Execution of arbitrary shell commands via the Bash tool during the evaluation judge phase. * Sanitization: No validation or sanitization of the commands or prompts within the YAML files is described.

agent-eval