The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes git commit messages and diffs, which are external inputs that could contain adversarial content designed to influence the agent's behavior during the review process.
Ingestion points: references/current-guidance.md instructs the agent to read commit messages, diffs, tests, and documentation.
Boundary markers: No specific delimiters are provided to isolate untrusted commit data from the agent's instructions.
Capability inventory: The skill is authorized to execute the review hook command to validate commits.
Sanitization: The instructions do not include steps to sanitize or filter the content of the commits before evaluation.
[COMMAND_EXECUTION]: The skill instructs the agent to execute a specific command-line tool for commit validation.
Evidence: references/current-guidance.md specifies the usage of review hook --review-scope commit-range --review-base <base> --review-head <head>.

commit-review