mcp-supabase-management

MCP server and LangChain tools for the Supabase Management API -- manage projects, databases, edge functions, secrets, and branches across all your Supabase organizations.

Features

24 tools across 7 categories:

Category	Tools	Description
Organizations	list_organizations	List all organizations
Projects	list_projects, get_project, create_project, delete_project, get_api_keys, get_project_url	Full project lifecycle management
Database	list_tables, execute_sql, list_extensions, list_migrations, generate_types	SQL execution, schema inspection, TypeScript type generation
Branches	list_branches, create_branch, delete_branch, reset_branch	Database branching for development workflows
Functions	list_functions, get_function, deploy_function	Edge function management and deployment
Secrets	list_secrets, set_secrets, delete_secrets	Project secret management
Logs	get_logs	Access logs for API, Auth, DB, Realtime, Storage, and Functions

Installation

# Core library
pip install .

# With MCP server support
pip install ".[mcp]"

# With LangChain tools
pip install ".[langchain]"

# Everything
pip install ".[all]"

Configuration

Variable	Required	Default	Description
SUPABASE_ACCESS_TOKEN	Yes	(none)	Supabase Management API access token (sbp_* prefix). Get one at supabase.com/dashboard/account/tokens

.env Example

SUPABASE_ACCESS_TOKEN=sbp_your_token_here

Quick Start

As MCP Server

mcp-supabase-management

Add to Claude Code (~/.claude.json):

{
  "mcpServers": {
    "supabase-management": {
      "type": "stdio",
      "command": "uv",
      "args": ["--directory", "/path/to/mcp-supabase-management", "run", "mcp-supabase-management"],
      "env": {
        "SUPABASE_ACCESS_TOKEN": "sbp_your_token_here"
      }
    }
  }
}

As LangChain Tools

from mcp_supabase_management.langchain_tools import TOOLS, supabase_list_projects, supabase_execute_sql

# Use all 24 tools with an agent
agent = create_react_agent(llm, TOOLS)

# Or use individual tools
print(supabase_list_projects.invoke({}))
print(supabase_execute_sql.invoke({
    "project_ref": "your_project_ref",
    "query": "SELECT * FROM users LIMIT 10",
}))

As Python Library

from mcp_supabase_management.client import SupabaseManagementClient
from mcp_supabase_management.operations import projects, database

client = SupabaseManagementClient(access_token="sbp_...")
all_projects = projects.list_projects(client)
result = database.execute_sql(client, "project_ref", "SELECT 1")

Error Handling

The server returns structured error messages:

{
  "error": "API Error (401): Invalid access token"
}

Common errors:

• 401 - Invalid or expired access token
• 404 - Project or resource not found
• 422 - Invalid request parameters
• 429 - Rate limit exceeded

License

MIT