thought-based-reasoning

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

• Third-party content exposure detected (high risk: 0.80). The ReAct section explicitly defines actions like Search[query] and Lookup[keyword] and shows examples (e.g., Search[High Plains]) where the agent ingests observations from external web/search results, which are arbitrary public third‑party content and could enable indirect prompt injection.