ai-fixing-errors

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly uses an external retriever (e.g., dspy.ColBERTv2(url="http://...") and dspy.Retrieve) and passes retrieved context.passages into the model (step3(context=context.passages,...)), which means it fetches and ingests potentially untrusted/public third‑party content as part of its workflow.