recipe-front-review
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository content through the
implementationFilesvariable and interpolates it into subagent prompts. This creates a surface where malicious instructions embedded in the codebase could influence the automated review and correction workflow. - Ingestion points: Step 1 identifies changed files (source code, tests, and configuration) to be reviewed.
- Boundary markers: Prompts for subagents in Steps 2, 3, 8, and 9 lack robust delimiters or specific instructions to ignore embedded directives within the analyzed files.
- Capability inventory: The workflow involves agents capable of performing security reviews, modifying implementation code, and updating design documentation.
- Sanitization: No explicit sanitization or validation of the untrusted file content is performed before it is processed by the LLM-based subagents.
Audit Metadata