requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use standard Git commands to identify commit ranges and generate code diffs.
- Evidence: SKILL.md contains instructions for using git rev-parse and git log to determine base and head SHAs.
- Evidence: code-reviewer.md includes git diff commands to allow the subagent to inspect code changes.
- [INDIRECT_PROMPT_INJECTION]: The skill uses a template for the code reviewer subagent that interpolates implementation details and plan requirements, creating a surface for indirect instructions.
- Ingestion points: The code-reviewer.md template accepts inputs for {WHAT_WAS_IMPLEMENTED}, {DESCRIPTION}, and {PLAN_REFERENCE} in SKILL.md and code-reviewer.md.
- Boundary markers: No specific delimiters are used to separate user-provided text from subagent instructions in code-reviewer.md.
- Capability inventory: The subagent is authorized to execute git diff commands in code-reviewer.md.
- Sanitization: No explicit sanitization or validation of the interpolated text is performed in the templates.
Audit Metadata