The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external code diffs and source files as primary inputs for its review process. This establishes an attack surface where maliciously crafted code could attempt to influence the reasoning of the reviewer subagents or the lead synthesizer. Ingestion points: Diff outputs from git diff and contents of files targeted for review are ingested into the prompt context. Boundary markers: Untrusted content is placed within specific sections defined by the {DIFF_OR_FILES} placeholder in the references/reviewer-prompt.md template. The skill does not explicitly define escaping or sanitization for these sections. Capability inventory: The agent can spawn subagents using the Task tool, read local files, execute git commands, and is instructed to perform repository writes by creating Pull Requests for configuration maintenance. Sanitization: No explicit content sanitization is described for the code being reviewed.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands like git diff to gather review context and suggests the automated creation of Pull Requests to update internal configuration files if model identifiers fail to resolve.

interrogate