The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data in the form of code changes. Maliciously crafted content or comments within the files being reviewed could attempt to mislead the agent or subvert the review process.
Ingestion points: File content read through standard tools and repository differences analyzed via git diff commands.
Boundary markers: The skill does not define specific isolation markers or instructions to help the model distinguish between its own instructions and instructions potentially embedded in the code under review.
Capability inventory: The agent has permissions for file system traversal (Read, Grep, Glob), shell execution (git, gh), and broad GitHub repository access via the mcp__github__ tools.
Sanitization: There are no explicit requirements for sanitizing or escaping code snippets before they are included in the final GitHub issue comment artifact.
[COMMAND_EXECUTION]: The skill utilizes shell commands such as git diff to identify changes and the gh CLI to post artifacts to GitHub. While these are standard developer actions, they represent direct interaction with the system shell.

comprehensive-review