how
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from the project codebase, creating an attack surface where instructions embedded in source files (such as in comments or documentation) could influence the agent's architectural synthesis.
- Ingestion points: Codebase content is ingested via the Read, Grep, and Glob tools as defined in the subagent prompt templates in the references directory.
- Boundary markers: The subagent prompts use structured Markdown and blockquotes to delineate findings but do not include specific instructions to disregard embedded commands or use specialized escaping.
- Capability inventory: The skill utilizes the Agent tool to spawn subagents restricted to readonly: true mode, and provides them with file discovery and reading capabilities.
- Sanitization: No automated filtering, sanitization, or content escaping is applied to the codebase data before it is interpolated into the prompts for synthesis.
Audit Metadata