XiYan MCP Server

Name: XiYan MCP Server
Author: XGenerationLab

@XGenerationLab

リポジトリ

XiYan MCP Server について

A Model Context Protocol (MCP) server that enables natural language queries to databases

#database #mcp #text-to-sql

基本情報

カテゴリ

その他

ライセンス

Apache-2.0

ランタイム

python

トランスポート

stdio

公開者

XGenerationLab

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

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

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is XiYan MCP Server?

A Model Context Protocol (MCP) server that enables natural language queries to databases. It is powered by XiYan-SQL, the state-of-the-art text-to-SQL system on open benchmarks, and is intended for developers and data analysts who want to query databases using natural language through MCP-compatible clients (e.g., Claude Desktop, Cursor).

How to use XiYan MCP Server?

Install the server with pip install xiyan-mcp-server, then provide a YAML configuration file specifying the LLM model, API key, database credentials, and transport protocol (stdio or sse). Launch the server with YML=path/to/yml python -m xiyan_mcp_server and connect it to a supported MCP client.

Key features of XiYan MCP Server

Query databases using natural language via XiYanSQL
Supports general LLMs (GPT, Qwen-max) and a SOTA text-to-SQL model
Runs in pure local mode for high security
Works with MySQL and PostgreSQL databases
Lists available tables and reads table contents as resources

Use cases of XiYan MCP Server

Build a local data assistant that answers questions from a database
Enable non-technical users to explore data via chat interfaces
Integrate natural-language database access into MCP-compatible IDEs or tools

FAQ from XiYan MCP Server

What databases does XiYan MCP Server support?

MySQL and PostgreSQL are currently supported. More dialects are planned.

Do I need an API key to use the server?

It depends on the model. General LLMs require an API key from the provider (e.g., OpenAI, Alibaba Cloud). The SOTA model can be used via Modelscope (requires a free API-inference key) or via Dashscope (key provided by email for trial). Local mode requires no API key.

Can I run the server entirely offline?

Yes. You can deploy the XiYanSQL-QwenCoder model locally and set the configuration to point to your local endpoint, achieving a fully local and high-security setup.