academic-pipeline

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

• Third-party content exposure detected (high risk: 0.90). The integrity_verification_agent (agents/integrity_verification_agent.md) explicitly requires WebSearch, DOI resolution, Semantic Scholar API checks and publisher/Google Scholar lookups for every reference and claim — untrusted public webpages are parsed and their results directly determine PASS/FAIL blocking decisions in the pipeline, creating a clear vector for indirect prompt injection via third‑party content.