SQLite toolkit build with FastMCP for LLM/agents frontends such as LM-Studio

FastMCP-based SQLite toolkit that can perform most standard SQL queries such as SELECT, INSERT, UPDATE, DELETE, ALTER, JOIN, ect'. Some of the simple or specific operation can performed via dedicated functions (tools) such as insert_row(), update_rows(), delete_rows(), add_column(), given explicit arguments. For complex queries or ones that are given entirely in natural language can be performed via raw SQL string with execute_sql_query(). Designed to be used with via MCP protocol with LLMs frontends such as LM Studio.

✅ Features

• Most standard SQL-style queries can be performed
• Both simple and complex queries are executed via dedicated functions or via raw SQL string
• Prefer the dedicated functions when possible — they’re safer and easier to use.
• Use execute_sql_query() for advanced cases or when the dedicated functions don’t fit.
• Input validation with Pydantic scheme
• Works seamlessly with LM Studio via MCP protocol

🚀 Installation

Install directly from GitHub:

pip install git+https://github.com/VadimDu/sqlite-toolkit-mcp.git

Or install in development mode:

git clone https://github.com/VadimDu/sqlite-toolkit-mcp.git
cd pdf-tool
pip install -e .

🛠 Testing

To test the this mcp-tool before real usage by LLM/agent:

python -m sqlite_tool.sqlite_tool_mcp_server

or

fastmcp run sqlite_tool/sqlite_tool_mcp_server.py

If everything is working fine, you should see a message like this: [date time] INFO Starting MCP server 'SQLite-tool' with transport 'stdio'

🎉 Usage in LM Studio

Add the following to your mcp.json file:

{
	"sqlite-tool": {
      "command": "python",
      "args": [
        "-m",
        "sqlite_tool.sqlite_tool_mcp_server"
      ]
    }
}