parallel-code-review
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external code diffs and pull request data, which represents a surface for indirect prompt injection if the code being reviewed contains malicious instructions.\n
- Ingestion points: The skill reads file contents and git diff outputs (SKILL.md).\n
- Boundary markers: The prompts include clear headers for subagent roles, but do not employ strict data/instruction delimiters or specific instructions to ignore content within the code blocks.\n
- Capability inventory: The instructions specify that subagents should be
readonly: true, which restricts their ability to modify the file system or environment based on injected instructions.\n - Sanitization: No explicit sanitization or filtering of the diff content is performed before processing.
Audit Metadata