flutter-genui

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

• Third-party content exposure detected (medium risk: 0.30). Outsider-authored text enters the runtime by being sent as part of Conversation.sendMessage(text) and then reflected back as untrusted A2UI JSONL streamed over SSE (A2uiTransportAdapter/TriageTransport.streamTriage) into SurfaceController.processMessage, which the renderer uses to resolve widget properties and function calls.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The code shows the client opening an SSE connection to an external agent endpoint (e.g., "https://api.your-app.com/triage/$jobId/stream" and "$baseUrl/triage/$jobId/stream") at runtime, and that SSE stream supplies A2UI JSONL messages which directly control the UI prompts/instructions; therefore this is a runtime external dependency that controls agent-driven prompts.