industrial-ai-research

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

• Third-party content exposure detected (high risk: 0.80). The skill's workflow (e.g., Phase 3 "Source Collection" and the Source Priority file) explicitly instructs the agent to gather and read public third‑party pages such as arXiv recent listings, IEEE Xplore landing pages, and publisher/conference pages and to use that content to drive search, triage, citation selection and synthesis, so it will ingest untrusted public web content that could carry indirect prompt injections.