The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code and test files to perform reviews as described in the Procedure section. This represents an attack surface for indirect prompt injection. The risk is mitigated by explicit boundary markers in the Scope section that require isolated execution and prohibit destructive actions. No specific sanitization of the ingested code is described.
[COMMAND_EXECUTION]: The skill is authorized to run local testing commands, specifically pytest, to validate software behavior as indicated in the evals/fixtures/review_project/CONTRACT.md and the evals/evals.json prompts. This capability is restricted to the local environment and is aligned with the primary purpose of test evaluation.

test-reviewer