crewai-agent-workflow

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

• Third-party content exposure detected (high risk: 0.70). Outsider free text can enter the LLM context via the Researcher’s retrieval tool (“retrieves source material … from the architecture-defined retrieval surface”) when that surface includes public web content fetched at runtime or other non-user-authored documents, whose readable text is then provided to the Researcher/Critic/Writer for synthesis and verification.