parallel-agents
Pass
Audited by Gen Agent Trust Hub on Apr 15, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill contains no executable code, scripts, or binary assets. It relies entirely on standard read-only tools (Read, Glob, Grep) to facilitate codebase discovery.
- [PROMPT_INJECTION]: The orchestration patterns describe passing findings from one agent to another (e.g., context passing and synthesis). This represents a surface for indirect prompt injection.
- Ingestion points: Local files read via Read, Glob, and Grep tools.
- Boundary markers: None identified in the synthesis templates or instructions.
- Capability inventory: Limited to read-only operations; no write or execution capabilities are present in the defined tools.
- Sanitization: No validation or escaping of agent outputs is documented before interpolation into subsequent prompts.
Audit Metadata