Neon MCP. Manage Postgres Infrastructure Via Conversation.

You gotta know this server. It lets your AI client handle every single piece of complex Postgres infrastructure without you even touching a dashboard. You run it straight from chat, treating your agent like a dedicated DBA that knows Neon inside and out.

🛠️ Setting Up Your Environment

When you need to scope out a whole new workspace, use create_project to set up an entirely scoped Neon project. This process automatically provisions the initial default resources, including necessary branches, databases, and read-write endpoints for that region and Postgres version. If you ever just need to change the name of the project—say, it was 'Test' but now it's 'Production v2'—you can use update_project to adjust only the human-readable label; this move won't mess with any underlying resources like your branches or databases.

To see what projects are available, you'll run list_projects, and if you need full metadata on one specific workspace, you fetch it using get_project. If things get messy and you gotta wipe the slate clean, delete_project vaporizes the entire Neon project, wiping out all associated branches, databases, and infrastructure—remember, the data is recoverable for seven days.

🌿 Branching and Database Management

For development work, list_branches shows you every isolated branch currently hanging around in your workspace. To get details on one specific branch, run get_branch. The core feature here is create_branch; this tool instantly clones a full, isolated PostgreSQL branch from any existing parent branch, giving you immediate isolation for whatever development task you're tackling.

If you need to designate the main source of truth for future cloning operations, use set_primary_branch.

When it comes to databases inside a project, you first list everything with list_databases, and if you need details on a specific branch, you can run get_connection_uri. To set up a new database within an existing Neon branch scope, simply use create_database and give it the name. For full visibility into what databases exist where, list_databases does the heavy lifting.

🔗 Connection Strings and Endpoints

Connecting your app is straightforward. Run get_connection_uri to grab a complete, ready-to-use PostgreSQL connection URI for any specified project or primary branch. For connectivity endpoints, first check what's available using list_endpoints. To provision compute access, use create_endpoint; this tool lets you set up either a read-write host—perfect for testing data changes—or a read-only host—ideal for analytics replication—for your selected Neon branch.

🔑 Security and Access Control

If you're dealing with credentials, the server handles it. To establish new user access, create_role generates a brand-new database user role and automatically creates its secure password, giving you ownership over those necessary resources. You can check who has rights by running list_roles, which displays all defined database user roles along with their specific privilege metadata within a branch.

These tools give your agent full control over securing access for development teams.

🗑️ Cleanup and Deletion

If you're done with a particular feature environment, don't just delete the data manually; use delete_branch to permanently destroy all associated resources from a specified Neon branch. Be careful here—that action is irreversible for the data itself. You can also list endpoints using list_endpoints and manage them via the other creation tools.

This server gives your agent full command over the entire Postgres lifecycle, letting you move from project scoping to connection URI generation without ever leaving the chat window.