Skip to content
Kitbay

neon-ai-gateway

Apache-2.0Updated on Nov 20, 2026117GitHub stars

A Neon skill for the AI Gateway built into each Neon branch, which lets an app call language, embedding and image models from several providers with one Neon credential instead of separate provider keys.

What the skill teaches

A Neon skill for the AI Gateway built into each Neon branch, which lets an app call language, embedding and image models from several providers with one Neon credential instead of separate provider keys. Each branch has its own gateway endpoint, so requests from a preview branch stay separate from production. The skill shows an agent how to enable the gateway, which environment variables to use, and how to point the Vercel AI SDK, Mastra, or the OpenAI, Anthropic and Google SDKs at it by changing the base URL. It also covers model identifiers, listing available models at runtime and setting the gateway up in neon.ts.

Capabilities

  • Call models from several providers through one Neon credential
  • Point the Vercel AI SDK or Mastra at the Neon AI Gateway
  • Generate embeddings and images through the gateway
  • List the models available on a branch