vercel-ai-sdk
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates a pattern where untrusted user input and external database content are interpolated directly into the system prompt for an LLM.
- Ingestion points: User-provided
messagesand database-retrievedproductsare ingested in theactionfunction withinapp/routes/api.chat.ts. - Boundary markers: The provided code examples do not use clear delimiters (such as XML tags or Markdown blocks) to separate untrusted content from system instructions.
- Capability inventory: The skill utilizes the
streamTextcapability to generate AI responses based on the provided context. - Sanitization: There is no evidence of sanitization or filtering of the
contextInfoormessagesbefore they are sent to the model. - [EXTERNAL_DOWNLOADS]: The skill instructs the user to install well-known and trusted packages from the Vercel ecosystem, specifically
aiand@ai-sdk/openai.
Audit Metadata