{"type":"mcp_client","name":"mcp-client-for-ollama","description":"Ollama MCP client: TUI for managing local LLMs. Multi-server, streaming, tool management, & full model config. Dev-focused.","category":"AI","language":"Python","stars":763,"forks":23,"owner":"jonigl","github_url":"https://github.com/jonigl/mcp-client-for-ollama","homepage":null,"setup":"## Setup\n\n### Prerequisites\n\nBefore installing `ollmcp`, ensure you have the following installed:\n\n*   **Python 3.10+**: You can download it from the [official Python website](https://www.python.org/downloads/).\n*   **Ollama**: Make sure Ollama is running locally. Download and install it from the [Ollama website](https://ollama.com/download).\n*   **UV package manager**: Install it using the instructions on the [UV GitHub repository](https://github.com/astral-sh/uv).\n\n### Installation\n\nYou can install `ollmcp` using one of the following methods:\n\n**Option 1: Install with pip**\n\n```bash\npip install --upgrade ollmcp\n```\n\nAfter installation, you can run the client using:\n\n```bash\nollmcp\n```\n\n**Option 2: Install with uv**\n\n```bash\nuvx ollmcp\n```\n\n**Option 3: Install from source (using a virtual environment)**\n\n1.  Clone the repository:\n\n    ```bash\n    git clone https://github.com/jonigl/mcp-client-for-ollama.git\n    ```\n\n2.  Navigate to the cloned directory:\n\n    ```bash\n    cd mcp-client-for-ollama\n    ```\n\n3.  Create and activate a virtual environment using `uv`:\n\n    ```bash\n    uv venv && source .venv/bin/activate\n    ```\n\n4.  Install the package and its dependencies:\n\n    ```bash\n    uv pip install .\n    ```\n\n5.  Run the client:\n\n    ```bash\n    uv run -m mcp_client_for_ollama\n    ```\n\n### Configuration\n\n*   **MCP Servers**: `ollmcp` can automatically discover MCP servers from Claude's configuration. Alternatively, you can specify server paths, URLs, or a JSON configuration file using command-line arguments (see \"Usage\" section).\n*   **Ollama Model**: The default model is `qwen2.5:7b`. You can change this using the `--model` or `-m` command-line argument.\n*   **Ollama Host**: The default host is `http://localhost:11434`. You can change this using the `--host` or `-H` command-line argument.\n*   **Configuration File**: `ollmcp` automatically loads the default configuration from `~/.config/ollmcp/config.json` if it exists.\n\n### Environment Variables\n\n`ollmcp` does not require any specific environment variables to be set. However, if your MCP servers require environment variables (e.g., API keys), you can set them in your shell environment or within the server configuration JSON (see \"Server Configuration Format\" section).","tools":"## Available Tools\n\n- 🌐 **Multi-Server Support**: Connect to multiple Model Context Protocol (MCP) servers simultaneously. This allows you to access a wider range of tools and capabilities within a single chat session.\n  *Example:* Connect to a weather server, a file system server, and a web search server all at once.\n\n- 🚀 **Multiple Transport Types**: Supports STDIO, SSE, and Streamable HTTP server connections. This provides flexibility in how you connect to MCP servers, accommodating different server implementations and network configurations.\n\n- 🎨 **Rich Terminal Interface**: Interactive console UI. Provides a user-friendly way to interact with the client and manage tools, models, and server connections.\n\n- 🌊 **Streaming Responses**: View model outputs in real-time as they're generated. This allows you to see the model's progress and get faster feedback.\n\n- 🛠️ **Tool Management**: Enable/disable specific tools or entire servers during chat sessions. This allows you to control which tools are available to the model, focusing on the capabilities you need.\n  *Example:* Disable the file system tool if you don't want the model to access your local files.\n\n- 🧑‍💻 **Human-in-the-Loop (HIL)**: Review and approve tool executions before they run for enhanced control and safety. This provides an additional layer of security and allows you to understand what tools the model wants to use and why.\n  *Example:* Before the model executes a command to delete a file, you can review the command and approve or deny it.\n\n- 🎮 **Advanced Model Configuration**: Fine-tune 10+ model parameters including temperature, sampling, repetition control, and more. This allows you to customize the model's behavior and optimize it for specific tasks.\n  *Example:* Adjust the temperature to control the randomness of the model's responses.\n\n- 💬 **System Prompt Customization**: Define and edit the system prompt to control model behavior and persona. This allows you to guide the model's responses and create a specific character or role.\n  *Example:* Set the system prompt to \"You are a helpful assistant that always provides accurate information.\"\n\n- 🎨 **Enhanced Tool Display**: Beautiful, structured visualization of tool executions with JSON syntax highlighting. Makes it easier to understand the tool calls and their results.\n\n- 🧠 **Context Management**: Control conversation memory with configurable retention settings. Allows you to manage the amount of conversation history that the model remembers.\n\n- 🤔 **Thinking Mode**: Advanced reasoning capabilities with visible thought processes for supported models (e.g., gpt-oss, deepseek-r1, qwen3, etc.). This allows you to see how the model is reasoning and understand its decision-making process.\n\n- 🗣️ **Cross-Language Support**: Seamlessly work with both Python and JavaScript MCP servers. This allows you to use tools written in different languages.\n\n- 🔍 **Auto-Discovery**: Automatically find and use Claude's existing MCP server configurations. Simplifies the process of connecting to MCP servers.\n\n- 🔁 **Dynamic Model Switching**: Switch between any installed Ollama model without restarting. Allows you to experiment with different models and find the best one for your needs.\n\n- 💾 **Configuration Persistence**: Save and load tool preferences between sessions. Saves time by allowing you to reuse your preferred settings.\n\n- 🔄 **Server Reloading**: Hot-reload MCP servers during development without restarting the client. Speeds up development by allowing you to quickly test changes to your MCP servers.\n\n- ✨ **Fuzzy Autocomplete**: Interactive, arrow-key command autocomplete with descriptions. Makes it easier to find and use commands.\n\n- 🏷️ **Dynamic Prompt**: Shows current model, thinking mode, and enabled tools. Provides context at a glance.\n\n- 📊 **Performance Metrics**: Detailed model performance data after each query, including duration timings and token counts. Allows you to monitor the model's performance and identify bottlenecks.\n\n- 🔌 **Plug-and-Play**: Works immediately with standard MCP-compliant tool servers. Simplifies the process of connecting to MCP servers.\n\n- 🔔 **Update Notifications**: Automatically detects when a new version is available. Keeps you up-to-date with the latest features and bug fixes.\n\n- 🖥️ **Modern CLI with Typer**: Grouped options, shell autocompletion, and improved help output. Provides a more user-friendly command-line experience.","faq":null,"created_at":"2025-04-23T21:05:02+00:00","updated_at":"2025-08-08T04:27:32+00:00","source_url":"https://model-context-protocol.com/clients/mcp-client-for-ollama","related_articles":[]}