requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The code-reviewer.md template interpolates {BASE_SHA} and {HEAD_SHA} directly into a bash command block. This presents a risk of command injection if these variables are populated with malicious shell metacharacters from untrusted sources rather than the output of git rev-parse as intended.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface in code-reviewer.md.
- Ingestion points: {WHAT_WAS_IMPLEMENTED}, {PLAN_OR_REQUIREMENTS}, and {DESCRIPTION} placeholders in code-reviewer.md.
- Boundary markers: None are used to delimit the external content from the subagent instructions.
- Capability inventory: The subagent has the capability to execute shell commands (git diff) and write assessment reports.
- Sanitization: No sanitization or validation of the interpolated data is performed before processing.
Audit Metadata