Dispatching Parallel Agents
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill describes a multi-agent workflow that processes code files, introducing a potential surface for indirect prompt injection.\n
- Ingestion points: Sub-agents ingest content from test files and source code directories (e.g.,
src/agents/agent-tool-abort.test.ts).\n - Boundary markers: The provided prompt templates do not utilize delimiters or specific instructions to disregard text-based commands found within the code files.\n
- Capability inventory: Dispatched agents are granted authority to modify source code and are encouraged to run tests to verify fixes.\n
- Sanitization: The instructions do not include steps to sanitize or filter the content of the files before the sub-agent processes it.
Audit Metadata