requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The instructions in SKILL.md and templates/code-reviewer.md require the agent to execute shell commands, specifically git, grep, and awk, to identify code changes and generate diffs for review.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the local git repository via git diff output which is then processed by a subagent. This creates an attack surface where malicious instructions could be embedded in the code or comments within the diff.
- Ingestion points: Git diff output from the repository.
- Boundary markers: Absent; no explicit instructions are provided to the subagent to ignore prompt-like content within the code diff.
- Capability inventory: Shell execution for git operations.
- Sanitization: Absent; the diff content is not validated or sanitized before being passed to the subagent.
- [DYNAMIC_EXECUTION]: The template in templates/code-reviewer.md performs runtime interpolation of shell variables ({BASE_SHA}, {HEAD_SHA}) into command strings.
Audit Metadata