requesting-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a subagent template that ingests external content, creating a risk for indirect prompt injection where malicious instructions in the reviewed code or plan could override agent behavior. * Ingestion points: Found in 'references/code-reviewer-template.md' via placeholders such as {WHAT_WAS_IMPLEMENTED}, {PLAN_OR_REQUIREMENTS}, and {DESCRIPTION}. * Boundary markers: The template uses Markdown headers and bold text to separate fields but does not include explicit instructions for the subagent to ignore commands within the provided data. * Capability inventory: The skill relies on 'git rev-parse' and 'git diff' commands. * Sanitization: There is no evidence of sanitization or filtering of the content populated into the placeholders.
- [COMMAND_EXECUTION]: The skill uses shell commands for Git operations, specifically 'git rev-parse' and 'git diff' in both 'SKILL.md' and the subagent template. These commands use placeholders ({BASE_SHA}, {HEAD_SHA}) that are intended to be Git commit identifiers. If the agent populates these with unsanitized data from a malicious user or environment, it could potentially lead to command injection.
Audit Metadata