Tinybird

🚀 Tinybird MCP Server

Tinybird MCP Server is a tool for data analysis and data pipeline management, enabling interaction with the Tinybird platform via the command - line interface. This document details its features, installation steps, configuration methods, and usage examples.

🚀 Quick Start

This guide will walk you through the installation, configuration, and usage of the Tinybird MCP Server.

✨ Features

Core Features

• Data Source Management: List and view data sources in the workspace.
• Pipeline Management: List and view pipeline endpoints in the workspace along with their SQL transformation logic.
• Data Request: Fetch data from pipeline endpoints via HTTP requests and support parameter filtering.
• Query Execution: Run SELECT queries directly on data sources to extract insights.
• Document Access: Provide built - in Tinybird product documentation for LLM to use as context information.
• Event Logging: Save user - generated prompts to a specified data source.
• Performance Analysis: Use the Tinybird analytics API to offer optimization suggestions for indexes, sorting keys, and partition keys.

Extended Features

• Data File Push: Combine with Filesystem MCP to create remote data sources or pipelines from local data files.

📦 Installation

Method 1: Using uvx

uvx install mcp-tinybird

Method 2: Manual Installation

1. Clone the repository:

   git clone https://github.com/modelcontextprotocol/servers.git
   cd servers/tinybird
   

2. Install dependencies and build:

   uv sync && uv build
   

Configuration

Create a .env file in the root directory and add the following content:

TB_API_URL=
TB_ADMIN_TOKEN=

Update the Claude Desktop configuration:

{
  "mcpServers": {
    "mcp-tinybird_local": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/your/mcp-tinybird",
        "run",
        "mcp-tinybird",
        "stdio"
      ]
    }
  }
}

💻 Usage Examples

Basic Usage

1. List data sources in the workspace:

   mcp-tinybird list data-sources
   

2. Get detailed information about a specific data source:

   mcp-tinybird get data-source "your_data_source_name"
   

Advanced Usage

3. Request data from a pipeline endpoint:

   mcp-tinybird request pipe "your_pipe_name" --params "param1=value1&param2=value2"
   

4. Execute a custom query:

   mcp-tinybird run-select-query "SELECT column1, COUNT(*) FROM your_table GROUP BY column1"
   

5. Analyze the performance of a pipeline endpoint:

   mcp-tinybird analyze pipe "your_pipe_name"
   

📚 Documentation

Project Structure

mcp-tinybird/
├── src/
│   ├── main.py          # Main program entry
│   ├── commands/       # Implementation modules for each functional command
│   └── resources/      # Resource files, such as documentation
└── README.md           # Project description document

Running and Debugging

1. Start in the development environment:

   uv run mcp-tinybird stdio
   

2. Debug using the MCP inspector:

   npx @modelcontextprotocol/inspector uv --directory /path/to/mcp-tinybird run mcp-tinybird
   

Packaging and Publishing

1. Sync dependencies and build:

   uv sync && uv build
   

2. Publish to PyPI:

   uv publish --token your_pypi_token
   

Monitoring and Optimization

To monitor the MCP server, you can integrate Prometheus and Grafana. Refer to the Monitoring Guide for specific methods.

For more detailed information, please visit the Tinybird Official Documentation or contact technical support.