M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
S

Supabase MCP Server

Official
Connect Supabase to your AI assistants
How to install ↓ Compare GitHub source repository ↗
Category
Database #3 of 58
Stars
★ 2.9k Very popular
Transport
stdio (local process) · Streamable HTTP
Runtime
Hosted (no local runtime) · Node.js
Credentials
API key / credential required
License
Apache-2.0
Last commit
74FMRS · B

Official server, feature-rich, but requires careful permission and risk management.

Strongest · Maintenance 18/20 Weakest · Reliability 10/20

Reliability
10/20
Security and permissions
15/20
Maintenance
18/20
Documentation
17/20
Setup experience
14/20
Why each score
Reliability 10/20
The repository shows a well-structured server with explicit feature groups and many tools. However, since this is a static analysis without executing code or inspecting tests/CI, we cannot confirm that the declared tools match real behavior. The static calibration rule caps reliability at 12 without verifiable execution evidence, and we deduct further for lack of evidence. Given the reputable maintainer and apparent organization, we assign 10.
Security and permissions 15/20
The server uses a personal access token for authentication and offers a read-only mode, which are good security practices. The license is Apache-2.0. However, we cannot verify from static files how tokens are handled, if dangerous operations require confirmation, or if data flows are fully disclosed. No red-line issues are apparent. We score 15 because security measures are evident but code audit is needed to confirm.
Maintenance 18/20
The repository is actively maintained with many stars and commits. The license is clear. Version 0.9.0 indicates regular releases. While we cannot verify issue response times or security channel, the maintenance signs are strong. We score 18.
Documentation 17/20
The README is comprehensive with setup steps, configuration examples, links to docs, and AI SDK usage. It covers installation, auth, features, and links to full tool docs. While limitations and troubleshooting are probably in linked docs, the provided documentation is high quality. We score 17.
Setup experience 14/20
The README provides clear setup instructions with client configuration examples and environment variable requirements. It supports both remote and local transports. Since we haven't executed the setup, we cannot be certain of smooth installation, but the documentation and package presence suggest it is straightforward. We score 14 due to lack of verifiable CI/test evidence for installation.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account dataConnects to a database

Best for

  • Developers and teams using Supabase
  • Users wanting to integrate AI into database workflows

Not for

  • Users not using Supabase
  • Scenarios requiring fully local and cloud-free operations (CLI has limited features)

Required permissions

  • Requires Supabase access token (SUPABASE_ACCESS_TOKEN)
  • May require OAuth 2.1 authorization (remote)
  • Read/write by default; read-only mode available (--read-only)

Risks and side effects

  • LLM may execute destructive operations such as dropping tables or deleting data
  • Token leakage could lead to unauthorized access
  • Remote server may log requests, posing privacy concerns

Setup

Before you start

Runtime：Hosted (no local runtime) · Node.js

SUPABASE_ACCESS_TOKEN requiredsecret Supabase personal access token for API authentication, generated from the Access Tokens page in your Supabase account settings; required when running the npm package.
  1. Read the security best practices.
  2. Configure the MCP client. For remote, use the URL https://mcp.supabase.com/mcp. For local CLI, use http://localhost:54321/mcp. The npm package can be run via npx with SUPABASE_ACCESS_TOKEN environment variable, and optional flags like --project-ref and --read-only.
claude_desktop_config.json 
{
  "mcpServers": {
    "supabase": {
      "type": "http",
      "url": "https://mcp.supabase.com/mcp"
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "supabase": {
      "type": "http",
      "url": "https://mcp.supabase.com/mcp"
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add --transport http supabase https://mcp.supabase.com/mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Tools such as get_project_url should appear in your MCP client's tool list; asking "What is my Supabase project URL?" and getting the real project URL confirms the connection.

Troubleshooting

  1. Ensure the access token is valid and not expired
  2. Verify project reference ID is correct
  3. Ensure the client supports HTTP transport
  4. For CLI, check that the local server is running

Things to try

Once connected, you can ask your AI assistant things like:

  • Get the URL of my Supabase project
  • List all tables in my project
  • Query the 10 most recent rows from the users table
  • Show me the current config of my project

Use cases

Let AI assistants manage Supabase tables
Query databases using natural language
Fetch project configuration and documentation

Supported clients

Claude Desktop
Cursor
Windsurf

Listed from the project's documentation, not tested by this site.

Overview

The Supabase MCP Server is an official MCP server that enables AI assistants like Cursor, Claude, and Windsurf to interact with the Supabase platform through the Model Context Protocol. It provides tools for managing tables, fetching configuration, querying data, and more. It supports both remote hosted and local CLI environments, with OAuth 2.1 for remote access and read-only mode options.

Similar servers

Compare with other MCP servers

Source revision 8a2650891e2f Data synced 2026-10-12 Read the FMRS scoring method