swarm-research

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.65). The required runtime workflow reads “worker outputs” (step 5: “Collect — Read all worker outputs when complete”) and then synthesizes them into an LLM context; since worker research can involve fetching/using public web content or other outsider-authored material, this creates an indirect prompt-injection exposure path via those outsider-generated research notes.