building-data-apps
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The integrated AI chat interface establishes a surface for indirect prompt injection attacks.
- Ingestion points: User-provided queries in the React frontend (
examples/react_chat_panel.jsx) and data retrieved from Google Cloud BigQuery tables. - Boundary markers: Present. The backend logic in
examples/express_chat.tsandexamples/fastapi_chat.pyutilizes system instructions and structured message formatting to define the model's operational scope. - Capability inventory: The skill initiates network requests to Google Cloud APIs and executes database queries within BigQuery.
- Sanitization: The application uses
react-markdownwithremark-gfmin the frontend to safely render responses, which helps prevent cross-site scripting (XSS) from potentially malicious data content.
Audit Metadata