Neon ships an official hosted MCP server at mcp.neon.tech that lets any AI model read and write your serverless Postgres databases, manage branching workflows, run migrations and inspect schemas — all through natural conversation. Authenticate with a Neon API key and connect in seconds.
https://mcp.neon.tech/sse
Claude Sonnet 4.5
MCP Playground runs 30+ models on the same workflow: switch anytime, or use Compare mode to run several in parallel and balance quality vs. cost.
Neon API key from console.neon.tech → Account Settings → API Keys. The key scopes access to all projects in your Neon account.
How models use it and what it is built for.
The Neon MCP server exposes Neon's platform API and SQL execution layer as MCP tools. Models can list projects and branches, run arbitrary SQL queries, create and delete database branches for isolated testing, apply schema migrations, inspect table structures and query row data. Neon's branching model makes it especially powerful for AI-driven development: the agent can create a branch, run a migration on it, verify the results, and then ask you to promote it — all in one conversation. The server is hosted by Neon, so there is nothing to run locally.
Typical tools an AI model can call. Exact names vary by version.
Copy any of these into MCP Agent Studio after connecting.
List all tables in the main branch and their row counts.
Create a new branch called "feature-pricing" and run this migration on it.
Show me the last 10 rows inserted into the "events" table.
What indexes exist on the "orders" table and are any missing for common queries?
This is not a single-model product: you get the same MCP connection with 30+ models (Claude, GPT, Gemini, DeepSeek, open-weight, and more), you can switch mid-conversation, and you can open Compare mode to run the same prompt against multiple models at once. The card above is a suggested starting point for this server — not the only choice.
Default pick for Neon
Claude Sonnet 4.5
Claude Sonnet 4.5 handles multi-step SQL reasoning and branching workflows reliably. Haiku is sufficient for simple data queries on smaller projects.
Open MCP Agent Studio with the connection pre-filled. Add your token, pick any of 30+ models, and start chatting — no install required.
Open Agent StudioCommon questions about connecting, scoping and using it safely.
It is an official hosted MCP server from Neon that lets AI models interact with your serverless Postgres databases. It covers SQL execution, schema inspection, branch management and migrations — all accessible from any MCP-compatible AI client.
Standard Postgres MCP servers connect to any Postgres instance via a connection string and focus on SQL queries. Neon MCP additionally exposes Neon's branching API — create isolated database branches for testing, run migrations safely and delete branches when done.
The MCP server is free to use. Neon has a generous free tier (0.5 GiB storage, unlimited branches). You pay for AI model usage separately — or use MCP Agent Studio's credit system.
Only if your database is hosted on Neon. For other Postgres providers (Supabase, Railway, self-hosted), use the generic Postgres MCP server instead.
Use a read-only Postgres role for query-only workflows. The Neon API key does grant write access at the platform level (branch creation). For production data, point the agent at a branch rather than the main branch.