The Agent Skills Directory

[SAFE]: The skill uses standard version control commands (git diff, git log, git rev-parse) to perform its primary function. It includes a validation step using git rev-parse to ensure the user-provided reference is valid before proceeding with data-heavy operations.- [PROMPT_INJECTION]: The skill is designed to process untrusted external data (git diffs, commit messages, and specification documents), which presents a surface for indirect prompt injection.
Ingestion points: Data is ingested via the output of git diff, git log, and the contents of files located in docs/, specs/, or .scratch/ directories.
Boundary markers: The sub-agent prompts do not explicitly define boundary markers or instruct the agents to ignore instructions embedded within the code or spec content.
Capability inventory: The skill has access to repository metadata and diffing tools. It uses parallel sub-agents to isolate the review axes, which limits the scope of any potential injection. It does not perform network operations or file-system writes.
Sanitization: There is no explicit sanitization of the diff content or spec text before it is passed to the sub-agents.

code-review