research-lookup

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

• Third-party content exposure detected (high risk: 0.85). The required workflow calls external research backends (Parallel Chat API / Perplexity via OpenRouter) and then injects the returned free-form research text (response.choices[0].message.content / response.choices[0].message.content) into the agent’s context as result["response"] (outsider-authored web/retrieved content).