AI SDK v6 Core
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Prompt Injection (LOW): The skill templates ingest untrusted data into LLM prompts, creating a surface for indirect prompt injection. \n
- Ingestion points: The
messagesarray intemplates/agent-with-tools.tsandtemplates/stream-text-chat.ts, and thepromptstring intemplates/generate-text-basic.ts. \n - Boundary markers: Absent; user input is passed directly to model components. \n
- Capability inventory: The agent in
templates/agent-with-tools.tshas tool-calling capabilities (getWeather,convertTemp,getAirQuality). \n - Sanitization: Absent in the templates; execution relies on the model's internal safety filters.\n- Data Exposure & Exfiltration (SAFE): No hardcoded secrets, sensitive file paths, or unauthorized data exfiltration patterns were found. Documentation explicitly advises on API key validation.\n- External Downloads (SAFE): All detected dependencies are standard, trusted packages from Vercel, Anthropic, and OpenAI.
Audit Metadata