understand
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase content which may contain malicious instructions designed to influence agent behavior or synthesis output.\n
- Ingestion points: Codebase files are ingested via search and read tools (grep, glob, read) as defined in the specialist prompts in SKILL.md.\n
- Boundary markers: The instructions for specialized subagents do not include delimiters or specific warnings to disregard instructions found within the analyzed source code.\n
- Capability inventory: The skill utilizes file system access tools and spawns multiple subagents to perform parallel analysis of the ingested data.\n
- Sanitization: No sanitization or validation of the codebase content is performed before the data is processed by the AI specialists.
Audit Metadata