mcp-maker 0.2.5

Auto-generate MCP servers from any data source. Zero code required.

⚒️ MCP-Maker

Auto-generate MCP servers from any data source. Zero code required.

Point MCP-Maker at a database, spreadsheet, or directory and get a fully functional MCP server in seconds — ready for Claude, ChatGPT, Cursor, and any MCP-compatible AI.

Note: MCP-Maker is built on top of the official MCP Python SDK (mcp package). The SDK is the low-level framework for building MCP servers in Python — you write every tool, resource, and handler yourself. MCP-Maker uses that SDK under the hood to auto-generate everything from your data source. Think of it like Django vs raw SQL: same power, less work.

---

How It Works

Your Data Source          MCP-Maker              MCP Server
┌──────────────┐    ┌──────────────────┐    ┌─────────────────────────┐
│ SQLite DB    │    │                  │    │ 📄 server.py (Editable) │
│ Google Sheet │───▶│  mcp-maker init  │───▶│   ↳ Add custom tools    │
│ Airtable     │    │                  │    │                         │
│ Notion DB    │    │  (auto-inspect)  │    │ ⚙️ _autogen_tools.py      │
│ CSV files    │    │  (auto-generate) │    │   ↳ list_users()        │
│ PostgreSQL   │    │                  │    │   ↳ search_users()      │
│ MySQL        │    └──────────────────┘    └─────────────────────────┘
└──────────────┘                            Ready for Claude ✅

🏢 Enterprise-Ready Architecture

• Non-Destructive Generation: mcp-maker init safely outputs two files: server.py (your editable entry point) and _autogen_tools.py. Regenerate as often as you like—your custom tools and business logic are never overwritten.
• Credential Hardening: Connection strings and API keys are strictly forbidden in generated files. They are securely loaded via generated .env files and standard environment variables.
• API Key Authentication (--auth api-key): Gate access to your generated server with an MCP_API_KEY environment variable. Every tool call is wrapped with an auth check.
• SSL/TLS by Default: All PostgreSQL and MySQL connections enforce encrypted transport (sslmode=require / ssl=True). Disable with --no-ssl for local development only.
• Schema Versioning & Migrations: Generates a .mcp-maker.lock file tracking your schema fingerprint. On re-generation, detects added/removed tables/columns and displays a detailed migration diff with color-coded changes, then auto-updates tools.
• Multi-Source Servers: Combine multiple data sources into a single server: mcp-maker init sqlite:///users.db mongodb://localhost/orders merges schemas and generates tools for both.
• Relationship Detection: Auto-discovers foreign keys in SQL databases and generates join_ tools (e.g., join_orders_with_users).
• Pagination Helpers: All list_ tools return {results, total, has_more, next_offset} for proper cursor-based pagination.
• Column Selection: Pass fields="name,email" to any list_ tool to receive only specific columns instead of SELECT *.
• Date Range Filters: When date/datetime columns are detected, list_ tools auto-generate date_from/date_to parameters.
• Batch Operations: batch_insert_ and batch_delete_ tools wrapped in transactions for all SQL connectors, MongoDB, and Supabase.
• Export Tools: export_{table}_csv() and export_{table}_json() for every table across all connectors.
• Webhook Support (--webhooks): Register, list, and remove webhooks for real-time notifications on insert/update/delete events.
• Redis Pub/Sub: publish_message(), channel_list(), and channel_subscribers() tools for Redis channels.
• Async Generation (--async): Generate async tools using aiosqlite, asyncpg, or aiomysql for high-concurrency MCP servers.
• LLM Context Optimization (--consolidate-threshold): For massive schemas (>20 tables), MCP-Maker intelligently switches from generating per-table CRUD tools to consolidated generic tools (e.g., query_database) to prevent LLM context window bloat and reasoning degradation.
• Structured Audit Logging (--audit): Optionally generate servers that output structured JSON logs for every tool invocation, ready for ingestion into Datadog, Splunk, or ELK.
• Async-Ready Connection Pooling: Fast, thread-safe database interactions with built-in connection pooling for PostgreSQL (psycopg2.pool), MySQL, and SQLite.
• Guaranteed Code Quality: All generated Python code is automatically syntactically verified through ast.parse and statically formatted with black.

✨ Why MCP-Maker?

• Zero-Code Generation: Instantly maps tables, columns, and APIs into atomic MCP tools. No coding required.
• Zero Vendor Lock-In: The generated mcp_server.py file is a 100% standalone, standard Python file built entirely on the official SDK. You can uninstall MCP-Maker immediately after generation and your server will continue to run flawlessly forever.
• Granular Security (--ops): You control exactly what the LLM can do. Explicitly authorize read, insert, update, or delete operations.
• Auto-Rate Limiting: Built-in TokenBucket throttling for Cloud APIs (Airtable, Notion, Google Sheets) prevents 429 Too Many Requests bans when LLMs make aggressive parallel tool calls.
• Semantic Vector Search (--semantic): Automatically spins up a ChromaDB vector database alongside your SQL tables for high-quality, meaning-based search.
• Context Window Optimized (--tables): Only expose the data you actually need to prevent LLM context bloat and hallucination.
• Environment Management: Safely manage API credentials via mcp-maker env instead of hardcoding them.

🔌 Supercharged Connectors

While basic tools stop at simple SQL tables, MCP-Maker's Connectors are engineered for complex enterprise use cases:

• HubSpot: Acts as an Enterprise Data-Bridge, intelligently auto-discovering custom objects, custom properties, mapping Deal Pipelines natively, resolving owner_ids to human names, and exposing batch upsert endpoints.
• Notion: Supports Multi-Database URIs (notion://DB1,DB2), automatically parses 20+ property types (Rollups, Relations, Formulas), handles cursor pagination, and exposes deep filter tools.
• Airtable: Generates tools to query via Airtable Formulas, target specific Views, automatically sorts records, and discovers your Bases directly (mcp-maker bases).
• Google Sheets: Treats entire tabs as discrete SQL-like tables, infers column types, and provides pinpoint update_cell tools.
• PostgreSQL / MySQL: Detects Primary Keys automatically, maps all complex SQL native types, and fully supports SSL-encrypted TLS connections.

Quick Start (2 minutes)

Step 1: Install

pip install mcp-maker

Step 2: Generate a server from your data

# SQLite database
mcp-maker init sqlite:///path/to/my_database.db

# CSV/JSON files in a directory
mcp-maker init ./my-data/

# Google Sheets (see docs for auth setup)
mcp-maker init gsheet://YOUR_SPREADSHEET_ID

# Airtable
mcp-maker init airtable://appXXXXXXXXXX

# Notion
mcp-maker init notion://DATABASE_ID

This creates a mcp_server.py file with all your tools.

Step 3: Connect to Claude Desktop

# Auto-configure Claude Desktop
mcp-maker config --install

# Restart Claude Desktop, then ask:
# "What tables are in my database?"
# "Show me the top 10 customers"
# "Search for records containing 'Python'"

That's it. Your AI can now query your data.

---

📖 Documentation

👉 Getting Started Guide — Tutorial, installation, first server, Claude Desktop setup

📋 CLI Reference & Architecture — CLI commands, env vars, security, architecture

Connector Guides (with step-by-step setup, examples, and troubleshooting)

Connector	Guide
SQLite	docs/sqlite.md
Files (CSV/JSON)	docs/files.md
PostgreSQL	docs/postgresql.md
MySQL	docs/mysql.md
Airtable	docs/airtable.md — Formulas, views, sorting, CRUD
Google Sheets	docs/google-sheets.md — GCP service account setup
Notion	docs/notion.md — Integration setup, multi-DB support
Excel	docs/excel.md — .xlsx files, sheets as tables
MongoDB	docs/mongodb.md — Document sampling, full CRUD
Supabase	docs/supabase.md — Auth/storage tools, PostgREST
REST API	docs/openapi.md — OpenAPI/Swagger specs
Redis	docs/redis.md — Key grouping, type-aware tools
HubSpot	docs/hubspot.md — Enterprise CRM Data-Bridge, PAT auth

Feature Guides

Feature	Guide
Semantic Search	docs/semantic-search.md — ChromaDB vector search, search by meaning

---

🔌 Supported Connectors (13)

Connector	URI Format	Auth Required	Install
SQLite	sqlite:///my.db	❌	Built-in
Files	./data/	❌	Built-in
PostgreSQL	postgres://user:pass@host/db	✅ DB creds	pip install "mcp-maker[postgres]"
MySQL	mysql://user:pass@host/db	✅ DB creds	pip install "mcp-maker[mysql]"
Airtable	airtable://appXXXX	✅ API key	pip install "mcp-maker[airtable]"
Google Sheets	gsheet://SPREADSHEET_ID	✅ Service acct	pip install "mcp-maker[gsheets]"
Notion	notion://DATABASE_ID	✅ Integration	pip install "mcp-maker[notion]"
Excel	./data.xlsx or excel:///path.xlsx	❌	pip install "mcp-maker[excel]"
MongoDB	mongodb://user:pass@host/db	✅ DB creds	pip install "mcp-maker[mongodb]"
Supabase	supabase://PROJECT_REF	✅ API key	pip install "mcp-maker[supabase]"
REST API	openapi:///path/to/spec.yaml	✅ API token	pip install "mcp-maker[openapi]"
Redis	redis://host:6379/0	✅ Password	pip install "mcp-maker[redis]"
HubSpot	hubspot://pat=YOUR_TOKEN	✅ PAT	pip install "mcp-maker[hubspot]"

Install all connectors at once:

pip install "mcp-maker[all]"

---

CLI Commands

mcp-maker init <source>                    # Generate an MCP server
mcp-maker init <source> --ops read,insert  # Include specific write operations
mcp-maker init <source> --tables users,orders  # Only include specific tables
mcp-maker init <source> --semantic         # Enable vector/semantic search
mcp-maker init <source> --audit            # Enable structured JSON audit logging
mcp-maker init <source> --auth api-key     # Require MCP_API_KEY for access
mcp-maker init <source> --async            # Generate async tools (aiosqlite/asyncpg)
mcp-maker init <source> --cache 60          # Cache read results for 60 seconds
mcp-maker init <source> --no-ssl           # Disable SSL for local development
mcp-maker init <source> --force            # Skip schema change warnings
mcp-maker init <source> --consolidate-threshold 10 # Consolidate large schemas
mcp-maker deploy --platform railway        # Generate deployment files
mcp-maker deploy --platform render         # Render deployment
mcp-maker deploy --platform fly            # Fly.io deployment
mcp-maker serve                            # Run the generated server
mcp-maker inspect <source>                 # Preview what would be generated (dry run)
mcp-maker config --install                 # Auto-write Claude Desktop config
mcp-maker env set KEY VALUE                # Store API keys safely in .env
mcp-maker env list                         # List stored keys (masked)
mcp-maker list-connectors                  # Show available connectors
mcp-maker bases                            # Discover Airtable bases

---

📦 Installation

# Core (SQLite + Files)
pip install mcp-maker

# With PostgreSQL support
pip install "mcp-maker[postgres]"

# With Airtable support
pip install "mcp-maker[airtable]"

# With Google Sheets support
pip install "mcp-maker[gsheets]"

# With Notion support
pip install "mcp-maker[notion]"

# With semantic search (ChromaDB vector search)
pip install "mcp-maker[semantic]"

# With async support
pip install "mcp-maker[async-sqlite]"    # aiosqlite
pip install "mcp-maker[async-postgres]"  # asyncpg
pip install "mcp-maker[async-mysql]"     # aiomysql

# New connectors
pip install "mcp-maker[excel]"           # Excel (.xlsx)
pip install "mcp-maker[mongodb]"         # MongoDB
pip install "mcp-maker[supabase]"        # Supabase
pip install "mcp-maker[openapi]"         # REST API (OpenAPI specs)
pip install "mcp-maker[redis]"           # Redis

# All connectors + semantic search + async
pip install "mcp-maker[all]"

# Development
pip install "mcp-maker[dev]"

---

🤝 Contributing

MCP-Maker is designed for community contributions — each new connector is a self-contained PR.

See CONTRIBUTING.md for detailed instructions and a step-by-step connector creation guide.

make install   # Set up dev environment
make check     # Run lint + tests

🔒 Security

Found a vulnerability? Please report it privately — see SECURITY.md.

📄 License

This project is licensed under the MIT License.

This project follows the Contributor Covenant Code of Conduct.