requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it instructs an agent to process and analyze external, untrusted code changes.
- Ingestion points: The
code-reviewer.mdtemplate ingests external data through the output ofgit diffcommands. - Boundary markers: The prompt uses Markdown headers and code blocks to structure the input, but it lacks explicit instructions for the sub-agent to ignore any natural language instructions that might be embedded within the code being reviewed.
- Capability inventory: The sub-agent is granted shell access to run Git commands (
git show,git log,git worktree) and provides technical evaluations that influence the primary agent's workflow. - Sanitization: There is no evidence of sanitization or filtering of the Git diff output before it is interpolated into the sub-agent's prompt context.
Audit Metadata