ai-sdk-ui-v7
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill templates, specifically in
templates/nextjs-api-route.ts, process untrusted user messages provided via a POST request. This is the intended purpose of the chat interface but represents a potential surface for indirect prompt injection. - Ingestion points: User input is ingested via
req.json()in the Next.js API route template. - Boundary markers: The templates use standard system instructions but do not include explicit delimiters to isolate user messages from the rest of the prompt context.
- Capability inventory: The skill enables the agent to stream text responses from an LLM (OpenAI) back to the user interface.
- Sanitization: No explicit input sanitization or output filtering is included in the templates, which is common for documentation-focused code samples.
Audit Metadata