The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code and diffs from the local environment and passes them to multiple LLM subagents for review. Maliciously crafted comments or code could attempt to influence the subagents' findings.
Ingestion points: SKILL.md (Step 1) identifies the scope of review and packages file contents and diffs.
Boundary markers: The prompt template in references/reviewer-prompt.md uses section headers like ## Intent and ## Code Under Review to separate user/code content from instructions.
Capability inventory: The skill utilizes the Task tool to spawn subagents and executes git diff for scope identification. Reviewer subagents are explicitly set to readonly: true.
Sanitization: No explicit sanitization or filtering of the code content is performed prior to analysis.
[COMMAND_EXECUTION]: The skill executes git diff to gather changes for the review process, which is an expected function for this tool.

interrogate