Tome

🚀 Tome Introduction

Tome is a magical tool for connecting local large language models (LLMs) via MCP servers.

🎉 Tome is Here

Tome is an application designed specifically for MacOS (support for Windows and Linux is coming soon). Developed by the Runebook team, it aims to help users connect their local LLMs to MCP servers. Tome manages your MCP servers, so you don't have to deal with uv/npm or JSON files. Simply connect it to Ollama, paste some MCP server configurations, and you can start chatting with MCP-powered models in seconds.

This is our first technical preview version, so be prepared to encounter some imperfections. Given the rapidly changing (and somewhat unstable) nature of MCP servers and local models, we recommend joining our Discord community to share tips, tricks, and any issues you encounter. Also, remember to star this repository on GitHub to stay informed about updates and new features.

🚀 Quick Start

Compatibility and Requirements

• MacOS (Sequoia 15.0 or higher is recommended).
• Ollama (either locally or remotely; you can configure any Ollama URL in the settings).

Download the latest version of Tome

Getting Started Quickly

We'll be updating our website in the coming weeks with more detailed documentation and step-by-step tutorials. Here's a quick interim guide:

1. Install Tome and Ollama.
2. Install a supported model (we recommend Qwen2.5, either the 14B or 7B version, depending on your memory).
3. Open the MCP tab in Tome and install your first MCP server (Fetch is a good starting point; just paste uvx mcp-server-fetch in the server field).
4. Start chatting with an MCP-powered model! Try getting the top stories from Hacker News.

🌟 Project Vision

Our goal is to make local large language models and MCP more accessible to everyone. We're building a tool that allows you to interact freely with LLMs in any situation, whether you're an engineer, a tech enthusiast, or an ordinary user.

Core Principles

• Tome is Local-First: You have control over where your data goes.
• Tome is for Everyone: You don't need to manage programming languages, package managers, or JSON configuration files.

Future Plans

• Model Support: Currently, Tome uses Ollama for model management, but we plan to expand support to other LLM engines and even cloud models. Let us know your requirements.
• Operating System Support: We plan to add support for Windows first, followed by Linux.
• More features are in development.

🤝 Join Us

Visit our GitHub repository for more information. Contact us via Issues to report problems or suggest improvements.