Vercel Connect now supports TanStack AI
Vercel Connect now lets agents built with TanStack AI call OAuth-protected servers without storing credentials.
Why it matters
Developer infrastructure for agentic applications just got easier: Vercel Connect adds native TanStack AI support, simplifying credential management and consent flows for agents calling external APIs. Practitioners building agent tooling on Vercel should evaluate this for production deployments.
The key facts
11 to knowVercel Connect adds @vercel/connect/tanstack-ai subpath export
OAuth-protected MCP servers now callable without credential storage or rotation
Fresh token generation on every MCP request via Connect-backed auth provider
Consent challenge handled before model execution (prevents consent errors reaching model as strings)
Supports TanStack transport config integration
Part of broader TanStack AI framework support in Vercel Connect
Vercel Connect adds TanStack AI support via @vercel/connect/tanstack-ai subpath export
OAuth token refresh happens automatically before every MCP request
Consent errors are caught before model execution, redirecting users rather than exposing errors to the model
New connectMCPTransport method simplifies integration with TanStack transport config
getConsentChallenge API enables proper redirect flow for unauthenticated requests
The story so far
Earlier coverage of this storyline
- Vercel Connect now supports Microsoft TeamsVercel Blog
- This story
Go to the source
Vercel Blogvercel.com
Publisher excerpt: Agents built with TanStack AI can now call OAuth-protected MCP servers through Vercel Connect, with no credentials for you to store or rotate. The new subpath exports , which takes a TanStack transport config and attaches a Connect-backed auth provider. The provider is called before every MCP…