The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to run a repository's test suite as part of the mutation testing process.
Evidence: In references/mutation-check.md, the procedure step 3 explicitly directs the agent to "Run the suite the repository defines, exactly as CI runs it."
Evidence: In SKILL.md, rule REVIEW-4 requires the agent to "run the suite" after inverting code conditions.
[DYNAMIC_EXECUTION]: The skill requires the agent to programmatically modify source code and execute the resulting changed logic at runtime.
Evidence: The mutation check procedure in references/mutation-check.md involves copying a file, inverting a condition (e.g., swapping operators or branches), and then executing the modified version.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface where the agent processes untrusted external data (pull request diffs and repository source code) that could contain malicious instructions.
Ingestion points: The agent is instructed to read PR intent, implementation files, and test files during "The walk" section of SKILL.md.
Boundary markers: There are no instructions for the agent to use specific delimiters or to disregard instructions embedded within the code comments or documentation it reviews.
Capability inventory: The agent possesses capabilities to read/write files and execute shell commands (test suites) based on its analysis of the PR.
Sanitization: The instructions do not prescribe sanitization or filtering of the ingested code/diffs before the agent processes them as part of the review task.

review-change