The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external agents and skills (via Read, Grep, and cat). This creates an attack surface where a malicious skill being evaluated could attempt to influence the evaluator's behavior through instructions embedded in its frontmatter or body. However, the skill mitigates this by using a deterministic scoring script (score-component.py) and a strict structural rubric.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard utility commands including ls, grep, wc, and python3. These commands are used locally to verify file existence, count lines, search for placeholders, and perform Python syntax checks (py_compile) on repository scripts.

agent-evaluation