AI导哈

Ollama 模型聊天与文件分析图形界面

Ollama MCP GUI 是一款桌面图形界面工具，面向需要本地大模型交互与文件内容分析的开发者，支持连接 Ollama、模型切换、聊天问答，并通过容器化 MCP 服务读取和分析文件。

  • by kaa-serpent
  • Python
  • 本地部署
  • 更新于
在 GitHub 查看 kaa-serpent/ollama-mcp-app

使用说明

Ollama MCP GUI

This project provides a graphical user interface (GUI) for interacting with local Ollama models and performing file analysis. It also includes a server component for handling file system operations in a containerized environment.

Features

  • Ollama Integration: Connect to a local Ollama instance to run large language models.
  • Model Management: List available Ollama models and switch between them.
  • Chat Interface: A simple chat interface to send prompts to Ollama models.
  • File Analysis: Read local files and use their content in prompts to be analyzed by Ollama.
  • Dockerized Server: The file operations are handled by a separate FastAPI server running in a Docker container.

Project Structure

  • main.py: The main PyQt5 GUI application.
  • ollama_manager.py: A class to manage interactions with the Ollama API.
  • mcp_client.py: A client to interact with the file operations server.
  • mcp-server/: Contains the FastAPI server for file operations.
    • mcp_server.py: The FastAPI application.
    • Dockerfile: Dockerfile for the server.
  • docker-compose.yml: Docker Compose file to build and run the mcp-server.
  • workspace/: A directory mounted into the Docker container for file operations.

Setup and Installation

Prerequisites

1. Install Python Dependencies

Install the required Python packages for the GUI client:

.\env\Scripts\activate

pip install -r requirements.txt

2. Build and Run the MCP Server

The file operation server runs in a Docker container. Use Docker Compose to build and start the server:

docker-compose up --build -d

This command will:

  • Build the Docker image for the mcp-server.
  • Start a container named mcp-server.
  • Mount the local ./workspace directory into the container at /app/workspace.
  • Expose the server on port 8000.

3. Run the GUI Application

Once the MCP server is running, you can start the GUI application:

python main.py

How to Use

  1. Start Ollama: Make sure your local Ollama instance is running.
  2. Start the MCP Server: Use docker-compose up --build -d.
  3. Launch the GUI: Run python main.py.
  4. Select a Model: The application will automatically load the available Ollama models. Select one from the dropdown menu.
  5. Chat with Ollama: Type your prompt in the chat input box and click "Send to Ollama".
  6. Analyze Files:
    • Place the files you want to analyze in the workspace directory.
    • Use the "File Tools" tab to read files directly.
    • In the "Chat & File Analysis" tab, you can reference a file in your prompt (e.g., "Summarize test.txt") and click "Analyze File with Ollama". The application will read the file and send its content along with your prompt to the selected Ollama model.

License

This project is licensed under the MIT License. See the LICENSE file for details.

相关 MCP