The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of source code and git diffs. Maliciously crafted comments or strings within the code being reviewed could attempt to manipulate the agent's behavior, review results, or verdict.
Ingestion points: Git diff output and the full content of modified files are read into the agent's context in Step 4.
Boundary markers: No specific delimiters or instructions are provided to the agent to treat the code content as untrusted or to ignore embedded instructions.
Capability inventory: The agent has the ability to read files, execute shell commands (git, gh), and modify files to apply fixes in Step 7.
Sanitization: No sanitization or filtering of the source code content is performed before analysis.
[COMMAND_EXECUTION]: The skill relies on executing shell commands to interact with the local environment and the GitHub API.
Evidence: Uses gh repo view to determine the default branch, git branch for the current branch, and git diff to extract code changes.

review-hamster