requesting-code-review

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard Git commands (git rev-parse, git log, git diff) to identify code changes and retrieve repository metadata. These operations are essential for the skill's primary purpose of code review.
  • [PROMPT_INJECTION]: The skill defines a 'Senior Code Reviewer' persona and provides structural instructions. This is a legitimate use of prompt engineering to focus the AI's behavior and does not involve bypassing safety filters or overriding system constraints.
  • [INDIRECT_PROMPT_INJECTION]: The code-reviewer.md template interpolates user-controlled variables ({DESCRIPTION}, {PLAN_OR_REQUIREMENTS}) directly into the agent's prompt.
  • Ingestion points: Content is injected into the reviewer agent's context through placeholders in code-reviewer.md.
  • Boundary markers: Absent. There are no explicit delimiters or instructions to the model to treat the content of the placeholders as untrusted data.
  • Capability inventory: The reviewer agent is intended to use Git tools to inspect the codebase.
  • Sanitization: No validation or escaping is applied to the interpolated text.
  • Note: While this creates an attack surface for indirect prompt injection if external requirements files are malicious, the risk is minimal given the intended developer-controlled environment and lack of high-privilege write/network capabilities.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 07:12 AM
Security Audit — agent-trust-hub — requesting-code-review