XiYan MCP Server for Natural Language Database Queries

Overview

The XiYan MCP Server is a Model Context Protocol (MCP) server designed to enable natural language queries to databases. Powered by XiYan-SQL, it provides state-of-the-art text-to-SQL capabilities, making it easier to interact with databases using plain language.

Key Features

• 🌐 Fetch data by natural language through XiYanSQL
• 🤖 Support for general LLMs (e.g., GPT, QwenMax) and Text-to-SQL SOTA models
• 💻 Pure local mode for high security
• 📝 Compatibility with MySQL and PostgreSQL
• 🖱️ List available tables as resources
• 🔧 Read table contents directly

Preview

Architecture

The server supports two integration modes:
1. Remote Mode: Requires an API key to access the XiYanSQL-QwenCoder-32B model from a service provider.
2. Local Mode: More secure, does not require an API key.

Best Practice

Learn how to build a local data assistant using MCP + Modelscope API-Inference without writing a single line of code: Best Practice Guide

Tools Preview

• get_data: Provides a natural language interface for retrieving data from a database.
• {dialect}://{table_name}: Retrieves sample data from the database for model reference.
• {dialect}://: Lists the names of the current databases.

Installation

Installing from pip

Python 3.11+ is required. Install the server via pip:

pip install xiyan-mcp-server

Run the server:

python -m xiyan_mcp_server

After configuring the YAML file, run the server with:

env YML=path/to/yml python -m xiyan_mcp_server

Installing from Smithery.ai

For installation via Smithery.ai, visit: @XGenerationLab/xiyan_mcp_server

Configuration

A YAML configuration file is required. Below is a sample config_demo.yml:

model:
  name: "XGenerationLab/XiYanSQL-QwenCoder-32B-2412"
  key: ""
  url: "https://api-inference.modelscope.cn/v1/"

database:
  host: "localhost"
  port: 3306
  user: "root"
  password: ""
  database: ""

LLM Configuration

Supported models include general LLMs, Text-to-SQL SOTA models, and local LLMs.

General LLMs

Example configuration for GPT-3.5:

model:
  name: "gpt-3.5-turbo"
  key: "YOUR_KEY"
  url: "https://api.openai.com/v1"

Text-to-SQL SOTA Model

For XiYanSQL-QwenCoder-32B, use:

model:
  name: "XGenerationLab/XiYanSQL-QwenCoder-32B-2412"
  key: "YOUR_KEY"
  url: "https://api-inference.modelscope.cn/v1/"

Local Model

To run the server in local mode:

1. Install additional Python packages:

pip install flask modelscope torch==2.2.2 accelerate>=0.26.0 numpy=2.2.3

1. Download the model manually:

modelscope download --model XGenerationLab/XiYanSQL-QwenCoder-3B-2502

1. Run the local server:

python local_xiyan_server.py

1. Configure the YAML file:

model:
  name: "xiyansql-qwencoder-3b"
  key: "KEY"
  url: "http://127.0.0.1:5090"

Database Configuration

Supported databases include MySQL and PostgreSQL.

MySQL

Example configuration:

database:
  host: "localhost"
  port: 3306
  user: "root"
  password: ""
  database: ""

PostgreSQL

Example configuration:

database:
  dialect: "postgresql"
  host: "localhost"
  port: 5432
  user: ""
  password: ""
  database: ""

Launch

Claude Desktop

Add the following to your Claude Desktop config file:

{
    "mcpServers": {
        "xiyan-mcp-server": {
            "command": "python",
            "args": ["-m", "xiyan_mcp_server"],
            "env": {
                "YML": "PATH/TO/YML"
            }
        }
    }
}

Cline

Prepare the config similar to Claude Desktop.

Goose

Add the following command in the config:

env YML=path/to/yml python -m xiyan_mcp_server

Cursor

Use the same command as Goose.

Witsy

Add the following in the command:

python -m xiyan_mcp_server

Set the environment variable YML to the path of your YAML file.

It Does Not Work!

For support, contact us via Ding Group or Weibo.

Citation

If you find our work helpful, please cite:

@article{xiyansql,
      title={A Preview of XiYan-SQL: A Multi-Generator Ensemble Framework for Text-to-SQL}, 
      author={Yingqi Gao and Yifu Liu and Xiaoxia Li and Xiaorong Shi and Yin Zhu and Yiming Wang and Shiqi Li and Wei Li and Yuntao Hong and Zhiling Luo and Jinyang Gao and Liyu Mou and Yu Li},
      year={2024},
      journal={arXiv preprint arXiv:2411.08599},
      url={https://arxiv.org/abs/2411.08599},
      primaryClass={cs.AI}
}

About

The XiYan MCP Server is a powerful tool for enabling natural language queries to databases, making database interactions more intuitive and efficient.

Topics

• Database
• MCP
• Text-to-SQL

Resources

• Readme

License

Apache-2.0 license