The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code, diffs, and pull requests which can contain malicious instructions intended to manipulate the agent. 1. Ingestion points: The agent reads file content and diffs using git and gh tools as described in Step 1. 2. Boundary markers: There are no instructions to use delimiters or ignore instructions embedded within the code under review. 3. Capability inventory: The agent is instructed to use shell tools like git and gh, and execute project-specific test and linting commands. 4. Sanitization: No sanitization of the input code is performed before analysis.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands for git operations and specifically to run the project's own linter and test commands. This presents a risk if the code being reviewed (e.g., in a pull request) has modified these scripts to perform malicious actions when executed.

code-reviewer