The Agent Skills Directory

[COMMAND_EXECUTION]: The skill describes the use of the Bash tool to interact with a CLI utility (agent-eval) and execute testing commands (e.g., pytest, npm run build) through its "judge" mechanism. This behavior is consistent with the skill's primary purpose of evaluating code generated by agents.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a surface where task definitions (YAML files) containing prompts and shell commands are processed by the agent.
Ingestion points: Task configuration files located in the tasks/ directory.
Boundary markers: None identified in the provided documentation.
Capability inventory: Uses Bash, Read, Write, and Edit tools to perform evaluations and record results.
Sanitization: No specific sanitization or validation logic for the YAML-defined commands is described in the markdown content.

agent-eval