dspy-miprov2-optimizer

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly retrieves and feeds remote, potentially untrusted text into the model (e.g., dspy.ColBERTv2(url='http://20.102.90.50:2017/wiki17_abstracts') and dspy.Retrieve(...) return passages that the agent reads/uses), exposing it to third‑party content that could enable indirect prompt injection.