waves

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow reads worker handoffs back into the orchestrator context, and those handoffs can include outsider-authored free text from web research sources (e.g., “vendor status pages and two community forums” in handoff-research.md) that the LLM then ingests during Step 3/3.5 synthesis.