The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted code as input for its review passes. This data ingestion surface is vulnerable to indirect prompt injection, where malicious instructions hidden in the code (e.g., comments or strings) could attempt to influence the agent's behavior or verdict. * Ingestion points: SKILL.md instructs the agent to read untracked files and git diffs using the Read and Bash tools. * Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions when reading source code. * Capability inventory: The skill utilizes Bash for command execution, Write for saving reports, and Agent for subagent delegation. * Sanitization: No specific sanitization or filtering of the read content is described.
[COMMAND_EXECUTION]: The skill allows for the execution of repository-specific test commands. While intended for verifying code correctness, this executes local code which could perform arbitrary actions if the repository environment is compromised. * Evidence: SKILL.md contains instructions to 'Run the repo's test command if one is obvious and cheap'.

reviewkit