deep-research

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow calls DeepAPI (curl ... /v1/research/deep), then parses .output.answer and .output.sources[].url and saves/uses that returned text as the agent’s research report, which can include outsider-authored free text from the web/other third parties.