swarm
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for fanning out tasks to multiple workers and aggregating their results, which creates a potential surface for instructions embedded in data to influence the agent.
- Ingestion points: Phase C (Aggregate) and Phase D (Report) instruct the agent to read terminal results and evidence from N workers.
- Boundary markers: The skill does not provide instructions for using delimiters or explicit boundary markers to isolate potentially untrusted worker output from the agent's aggregation logic.
- Capability inventory: The skill utilizes the host's task facility to launch concurrent workers and then generates a consolidated report based on the collected evidence.
- Sanitization: There is no mention of sanitization, validation, or escaping of the data collected from workers before it is processed and reported in the final chat consolidated report.
Audit Metadata