logos-router-reasoning-mesh

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

• Third-party content exposure detected (medium risk: 0.65). The required runtime workflow can ingest arbitrary outsider-authored free text from user-provided inputs (e.g., router.query(prompt=...) / workflow input_text / document analysis content), and that text is then sent into the router’s LLM context for reasoning/synthesis.