GraceYang9127/MCP_server_stdio

MCP_server_stdio

FastMCP Server (STDIO) with File, System, and Audio Tools

This project is a demonstration of a FastMCP server that uses the stdio transport, exposing several useful tools for:

• File operations: read_file, list_files, rename_file
• Host system information: get_host_info
• Audio transcription via OpenAI Whisper: transcribe_audio

Prerequisites

• Python 3.12
• ffmpeg installed and available on your system path
• openai-whisper package installed
• python-dotenv package installed
• Either conda or venv for environment isolation (recommended)

Setup

1. Create a new Conda environment (recommended)

conda create -n mcp_stdio_env python=3.12
conda activate mcp_stdio_env

2. Install dependencies

pip install -r requirements.txt

3. Testing

3. Configure file path with .env

• Create a file named. .env in the project root

MCP_BASE_DIR=/full/path/to/your/test/folder

• The tools expect your files to live in a subdirectory named test/.
• Make sure it exists and contains the files you want to work with.
• For audio transcription, place audio files here.

Usage

1. Start the server

python myServer.py

• This will run a FastMCP server using stdio transport. The server will stay active and wait for commands from a client.

2. Connect a client

• You’ll need a FastMCP-compatible client (or you can build your own).
• Tools are exposed automatically and follow the schema defined in tools.py.