Tulip MCP Server

by tulip

Not rated

GitHub

About

An MCP server for the Tulip API, allowing LLMs to interact with the Tulip manufacturing platform's tables, records, machines, and more.

Details

Author: tulip
Categories: Developer Tools, Other, Database, API

Multiple Workspace Configuration (Enterprise)

If your organization uses multiple Tulip workspaces or instances, you can set up multiple MCP servers to access all of them simultaneously. This lets you work with data from all your workspaces in a single conversation.

Before you begin, check what type of API credentials you have:

Workspace API Credentials: Created inWorkspace Settings→API Tokens

- ✅ Already know which workspace they belong to
- ✅Do NOT includeTULIP_WORKSPACE_IDin your.envfile
- ✅ Most common type for individual workspaces

Account API Credentials: Created inAccount Settings→API Tokens

- ⚠️ Can access multiple workspaces
- ⚠️Must includeTULIP_WORKSPACE_IDin your.envfile

Not sure which type you have?Check where you created your API token. If you created it in Workspace Settings, you have Workspace API credentials.

Step 1: Create separate.envfiles for each workspace

Follow the same process from](https://modelcontextprotocol.io/quickstart/user)Section 1: Configure Your Credentials, but create separate files:

production-workspace.env development-workspace.env

For each workspace, create a.envfile with the appropriate credentials:

# production-workspace.env TULIP_API_KEY=your_production_workspace_api_key TULIP_API_SECRET=your_production_workspace_secret TULIP_BASE_URL=https://your-instance.tulip.co ENABLED_TOOLS=read-only,table,station

# production-workspace.env TULIP_API_KEY=your_account_api_key TULIP_API_SECRET=your_account_secret TULIP_BASE_URL=https://your-instance.tulip.co TULIP_WORKSPACE_ID=PRODUCTION_WORKSPACE_ID ENABLED_TOOLS=read-only,table,station

Step 3: Connect multiple servers to your MCP client

Add each workspace as a separate server with a unique name using the guides from Section: Connecting to an MCP Client:

{ "mcpServers": { "tulip-production": { "command": "npx", "args": ["@tulip/mcp-server", "--env", "/full/path/to/production-workspace.env"] }, "tulip-qa": { "command": "npx", "args": ["@tulip/mcp-server", "--env", "/full/path/to/qa-workspace.env"] } } }

For Cursor:Use the install button multiple times, once for each.envfile.

- Use clear server namesliketulip-production,tulip-qa,tulip-development
- Test each workspace separatelyfirst to ensure credentials work
- Only enable the tools you need. Enabling too many tools (40+) can confuse the AI.

For detailed tool documentation including complete parameter lists, examples, and required permissions,generate theTOOLS.mdfile by runningnpm run docs.
- Log in to your Tulip instance.
- Navigate toSettings>API Tokens.
- Create a new API token. Give it a name (e.g., "MCP Server").
- Make sure to grant it the necessary permissions (scopes). A good starting set for read-only access is:stations:read,users:read,tables:read,machines:read,apps:read,urls:sign
- Copy theAPI KeyandSecretand paste them into your.envfile.

⚠️Important: TheTULIP_WORKSPACE_IDis only required if you are using an Account API key (obtained from Account Settings). If you are using a Workspace API key (obtained from Workspace Settings), you can leave this field empty.

This is a web browser that enables your coding agent, such as Claude Code, to visit websites on your behalf and assist you in identifying bugs or creating UI test cases.

Hosted MCP server and coordination layer for AI coding agents — live API contracts, database schema, frontend/backend mismatch detection, and shared handoff tickets for Claude Code, Cursor, Codex, and Lovable.

This server enables AI assistants and other MCP clients to interact with Directus instances programmatically.

An MCP server for the Frappe Framework, enabling AI assistants to interact with Frappe's REST API for document management and schema operations.

A FastAPI microservice for health-related operations, featuring JWT authentication and a PostgreSQL database with Alembic migrations.

A GraphQL server that supports the Model Context Protocol (MCP), enabling Large Language Models (LLMs) to interact with GraphQL APIs through schema introspection and query execution.

Persistent, inspectable memory for AI agents via hosted MCP and API. Supports recall, structured query, lineage, correction, and tenant-scoped remote memory.

Interact with Gibson projects to create/update projects, explain database/API interactions, and write code within your IDE.

Interact with Odoo instances using the XML-RPC API. Requires configuration via environment variables or config files.

Control Opentrons robots, manage protocols, and search API documentation.

An MCP server for interfacing with SIEMENS WinCC Unified SCADA systems via their GraphQL API.

An MCP server for the Tulip API, allowing LLMs to interact with the Tulip manufacturing platform's tables, records, machines, and more.

A Model Context Protocol (MCP) server that provides comprehensive access to the Tulip API, enabling LLMs to interact with the Tulip manufacturing platform functionality including tables, records, machines, stations, interfaces, users, and more.

Before you begin, ensure you haveNode.jsinstalled on your system. This is required to run the server.

This guide will walk you through running the server and connecting it to an MCP client like Cursor or Claude Desktop.

Create a file named.envin a folder of your choice. Copy and paste the following, replacing the placeholders with your actual Tulip credentials.

- YourTULIP_BASE_URLis the URL you use to access Tulip (e.g.,https://my-company.tulip.co).
- YourTULIP_WORKSPACE_IDis in your Tulip URL after/w/(for most users, this isDEFAULT).

TULIP_API_KEY=your_api_key_here TULIP_API_SECRET=your_api_secret_here TULIP_BASE_URL=https://your-instance.tulip.co TULIP_WORKSPACE_ID=your_workspace_id_here_if_using_account_api_key

Open your terminal or command prompt, navigate to the folder containing your.envfile, and run:

The server will start and is now ready to be connected to an MCP client.

When using a client, the server is run in a different environment where it may not find your.envfile automatically. To solve this, you must provide the full path to your.envfile using the--envflag.
- Navigate to the folder where you created your.envfile.
- On Windows:Right-click the.envfile while holding down theShiftkey, then select"Copy as path".
- On macOS:Right-click the.envfile, hold down theOptionkey, then select"Copy .env as Pathname".
- You will use this copied path in the client configuration below.
- From the Claude Desktop menu bar, selectSettings...>Developer>Edit Config.
- This will open theclaude_desktop_config.jsonfile.
- Add the server configuration inside themcpServersobject.You must replace"C:\\path\\to\\your\\.env"with the actual path you copied.

{ "mcpServers": { "tulip-mcp": { "command": "npx", "args": [ "@tulip/mcp-server", "--env", "C:\\path\\to\\your\\.env" ] } } }

For more details, see theofficial Claude Desktop MCP Quickstart.

For the easiest setup, click the button below. This will pre-fill the command.

After clicking the button,you must replace the placeholder text(REPLACE_WITH_YOUR_ENV_FILE_PATH_HERE) with the full path to your.envfile that you copied earlier.

This section contains more advanced configuration features.

By default, the server enables onlyread-onlytools andtabletools for safety. You can customize which tools are available using theENABLED_TOOLSenvironment variable in your.envfile.

TheENABLED_TOOLSvariable accepts a comma-separated list that can include:

- Individual tool names: Specific tools likelistStations
- Categories: Security-based groupings (read-only,write,admin)
- Types: Resource-based groupings (table,machine,user,app,interface,station,station-group,utility)

# Enable specific tools only ENABLED_TOOLS=listTables,getTable,listStations,listInterfaces # Enable by security category ENABLED_TOOLS=read-only,write # Enable by resource type ENABLED_TOOLS=table,station,interface # Mixed approach (recommended) ENABLED_TOOLS=read-only,interface,station,user # Enable everything (use with caution) ENABLED_TOOLS=read-only,write,admin