atlascloud

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

• Third-party content exposure detected (high risk: 0.75). The required workflow is calling Atlas Cloud HTTP endpoints (chat/image/video/poll/upload) and ingesting the returned JSON fields like .choices[0].message.content or .data.outputs into the agent’s context; these responses are generated by an external third-party service (outsider source: public API responses fetched at runtime).