Xata MCP server

- A hosted MCP server that runs alongside the Xata API — there is nothing to install or run locally.
- Authenticated via OAuth in your browser, or with a Xata API key for headless environments.
- Accessible from any MCP client that supports remote servers over Streamable HTTP.

The server uses theStreamable HTTPtransport. There is no SSE endpoint and no local (npm) version of the server.

The MCP server supports two authentication methods:

Clients that support custom headers can authenticate with aXata API keyinstead:

Create a dedicated API key for MCP access rather than reusing an existing key. Store it in an environment variable or your client’s secret storage — never commit it to source control.

Cursor offers a deep link for quick OAuth setup:Add to Cursor
- Open the command palette and search for “Cursor Settings”.
- UnderTools & MCP, clickNew MCP Server.
- Add the Xata server to the configuration file that opens:

{ "mcpServers": { "xata": { "url": "https://api.xata.tech/mcp" } } }

- Save the file. Cursor prompts you to authenticate — follow the browser flow and approve access to your Xata account.

claude mcp add --transport http xata https://api.xata.tech/mcp

Then start Claude Code and run the/mcpslash command. Select thexataserver and follow the browser instructions to authenticate. To use an API key instead of OAuth (for example, in CI):

claude mcp add --transport http xata https://api.xata.tech/mcp --header "Authorization: Bearer YOUR_XATA_API_KEY"

Alternatively, add it to your configuration manually:

{ "servers": { "xata": { "type": "http", "url": "https://api.xata.tech/mcp" } } }

Start the server fromMCP: List Serversand allow it to authenticate when prompted.

Connect ChatGPT to Xata using a custom connector:
- In ChatGPT, go toSettings→Connectors→Advanced settingsand enableDeveloper mode.
- On the Connectors tab, create a new connector with the server URL:
- ChooseOAuthfor authentication and complete the authorization flow when prompted.
- In each chat where you want to use Xata, click the+button and enable the Xata connector underAdd sources.

codex mcp add xata --url https://api.xata.tech/mcp

Theaddcommand may open a browser and report an OAuth error. If that happens, continue with the login command below; thexataserver entry has already been saved.

Authenticate with Xata using explicit OAuth scopes:

codex mcp login xata --scopes mcp-client,offline_access

Complete authorization in the browser. Theoffline_accessscope allows Codex to refresh its Xata session without requiring another browser authorization. Then startcodex, run/mcp, and verify thatxatais connected and authenticated.

Add Xata to your global MCP configuration:

{ "mcpServers": { "xata": { "serverUrl": "https://api.xata.tech/mcp" } } }

To enable Xata only for one project, use.agents/mcp_config.jsonin that project’s root instead. Startagyand enter/mcp. In the MCP Manager, useAuthenticateforxataand follow the prompts to complete OAuth.

Add the Xata server to your OpenCode configuration file:

{ "$schema": "https://opencode.ai/config.json", "mcp": { "xata": { "type": "remote", "url": "https://api.xata.tech/mcp" } } }

amp mcp add xata https://api.xata.tech/mcp

Then startamp— you should be prompted to authenticate in the browser. Run/mcp list toolsto confirm the server is connected.
- In Windsurf, open the Cascade panel and click the MCP (hammer) icon, thenConfigureto open the raw configuration file (~/.codeium/windsurf/mcp_config.json).
- Add the Xata server entry:

{ "mcpServers": { "xata": { "serverUrl": "https://api.xata.tech/mcp" } } }

- Save the file and clickRefreshin the Cascade sidebar. Complete the OAuth flow when the browser window opens.
- OpenSettings→AI→MCP Serversand clickAdd Server→Add Remote Server, or edit your settings file directly:

{ "context_servers": { "xata": { "url": "https://api.xata.tech/mcp" } } }

- Zed prompts you to authenticate against the server using the standard MCP OAuth flow.
- Open Cline in VS Code and click theMCP Serversicon.
- In theRemote Serverstab, enterxataas the name,https://api.xata.tech/mcpas the URL, and chooseStreamable HTTPas the transport. Or edit the configuration JSON directly:

{ "mcpServers": { "xata": { "type": "streamableHttp", "url": "https://api.xata.tech/mcp" } } }

The transport type must bestreamableHttp(camelCase). Omitting it causes Cline to fall back to the legacy SSE transport, which the Xata MCP server does not support.

Any MCP client can connect if it supports:

- Remote MCP servers overStreamable HTTP(not SSE)
- OAuth with dynamic client registration, orcustom HTTP headersfor API key authentication

Consult your client’s documentation for where to configure remote MCP servers, and usehttps://api.xata.tech/mcpas the URL.

Use the Xata MCP server to tell me who I’m authenticated as.

The assistant should call theuser_infotool and return your user identity (or the API key identity, if you authenticated with a key). If it does, the connection is working.

The Xata MCP server exposes the following tools:

- Prefer OAuth for interactive clients; tokens are short-lived and can be revoked by disconnecting the server in your client.
- For automation, use a dedicatedAPI keyand rotate it regularly.
- Some tools can modify your data:call_write_operationandcall_destructive_operationcan change or delete resources (the latter requiresconfirm=true), andrun_sqlcan mutate data when called withwrite=true. Review the actions your assistant proposes before approving them, and keep a human in the loop for any write or delete.

Authentication keeps failing or loops.Remove the Xata server from your client, restart the client, and add the server again to trigger a fresh OAuth flow.The server connects but no tools show up.Make sure you completed the authentication step — most tools require a valid session before they appear. Re-run your client’s authentication flow, then refresh its tool list. See Available tools for the full set.Your client can’t connect at all.Confirm the URL is exactlyhttps://api.xata.tech/mcpand that your client supports Streamable HTTP. SSE-only clients are not supported.The server doesn’t appear in your client.Check the client’s MCP configuration file syntax — the JSON shape differs between clients (mcpServersvsserversvscontext_servers,urlvsserverUrl) — and check the client’s logs. Most clients require a full restart after configuration changes.

Postgres skills and documentation to help AI coding tools generate better PostgreSQL code.

Build custom MCP tools on any datasource and ship them to any agent builder from one control plane—using only SQL and a secure link.

Hosted MCP server for live sports data — scores, analytics, schedules, standings, multi-book odds, team form, head-to-head, model predictions, and pre-generated matchup analysis across 1,000+ leagues in 150+ countries. Free tier, no card.

A secure MCP server that connects AI agents to CloudBeaver web client, allowing them to explore database schemas, read metadata, and execute SQL queries.

Real-time company and people intelligence for AI agents

Fabi MCP is an autonomous agent that handles end-to-end data analysis tasks from natural language requests, automatically discovering data schemas, generating sql or python code, executing queries, and presenting insights.

An enterprise-ready system to archive AI conversations from ChatGPT and Claude into a Supabase database.