Focus SQL

🚀 FOCUS Dataset MCP Server

The Model Context Protocol (MCP) server enables AI assistants to convert natural language into SQL statements.

🤔 Why Another Text-to-SQL Framework?

In short, focus_mcp_sql adopts a two-step SQL generation solution, which controls LLM hallucinations and ensures output standardization. Compared with other frameworks, it has obvious advantages in functionality and performance.

Here is a comparative analysis with existing frameworks:

✨ Features

Model Initialization

Initialize a domain-specific language model by providing dataset metadata, including table structures, field attributes, and inter-table relationships.

Natural Language to SQL Conversion

Based on the initialized model, convert user-input natural language queries into standard-compliant SQL statements.

📦 Prerequisites

Installation Requirements

• Operating System: Linux, Windows
• Memory: >=4GB
• Disk Space: >=10GB

Dependencies

• Java 8+
• Python 3.7+

📦 Installation

1. Download the Code Repository

git clone https://github.com/your-organization/focus-mcp.git
cd focus-mcp

2. Environment Configuration

Create a config.json file with the following content:

{
  "model_path": "path/to/model",
  "data_source": "path/to/data"
}

3. Start the Service

python server.py

🔧 MCP Configuration

Initialize the Model (tool: gpt_text2sql_init)

Parameter Description

• model_id: A unique identifier to distinguish context information of different datasets.
• schema: Data table structure, including table names, field names, and field types.
• properties: Field attributes, including constraints such as primary keys and foreign keys.
• relations: Inter-table relationships, including relationship types (inner join, outer join) and association conditions.

Example

{
  "model_id": "your-model-id",
  "schema": {
    "tables": [
      {
        "name": "employees",
        "columns": ["id", "name", "age"]
      },
      {
        "name": "departments",
        "columns": ["did", "dname"]
      }
    ]
  },
  "properties": {
    "id": { "primaryKey": true },
    "did": { "foreignKey": "employees.id" }
  },
  "relations": [
    {
      "type": "inner-join",
      "condition": "employees.did = departments.did"
    }
  ]
}

Natural Language to SQL Conversion (tool: gpt_text2sql_chat)

Parameter Description

• chat_id: A unique identifier for the conversation.
• query: The user-input natural language query.
• context: Context information, including the initialized model ID and associated table information.

Example

{
  "chat_id": "your-chat-id",
  "query": "Show the maximum age of employees",
  "context": {
    "model_id": "your-model-id",
    "tables": ["employees"]
  }
}

📄 Response Format

The return results of all tools follow the following format:

{
  "errCode": 0,
  "exception": "",
  "msgParams": null,
  "promptMsg": null,
  "success": true,
  "data": {
    // Specific business data
  }
}

💻 Usage Examples in VS Code

1. Install the C++ Visual Debugging Tool

2. Configure the MCP Server

3. Usage

• Initialize the model
• Execute a query: Show the maximum age of employees

📞 Contact Information

https://discord.gg/AVufPnpaad