Dive is an open-source desktop application that acts as a Model Context Protocol (MCP) host, enabling seamless integration with any Large Language Model (LLM) that supports function calling. It targets developers and power users looking to build sophisticated AI agents that can interact with external tools and services, offering universal LLM compatibility and cross-platform support.

How It Works

Dive leverages the Model Context Protocol (MCP) to facilitate communication between LLMs and external tools. It acts as a host, managing the execution of functions called by the LLM and returning results. The application supports both standard input/output (stdio) and Server-Sent Events (SSE) for MCP integration, allowing flexibility in how LLMs connect and interact with the platform.

Quick Start & Requirements

• Installation: Download the appropriate executable (.exe, .dmg, .AppImage) for your OS from the releases page.
• Prerequisites:
  • Windows: Python and Node.js are pre-installed.
  • macOS/Linux: Requires manual installation of Python and Node.js (with npx uvx).
  • yt-dlp-mcp requires yt-dlp to be installed separately (e.g., pip install yt-dlp).
• Setup: Download and run the executable. Configuration for tools like fetch, filesystem, and youtubedl can be added via JSON.
• Links: Releases, MCP Server Configuration

Highlighted Details

• Universal LLM support including ChatGPT, Anthropic, Ollama, Google Gemini, and Mistral AI.
• Cross-platform compatibility for Windows, macOS, and Linux.
• Advanced API management with support for multiple API keys and model switching.
• Includes a mechanism for automatic application updates.

Maintenance & Community

• The project is actively maintained, with recent updates including a rewrite of the host application in Python.
• Community channels include Discord and Twitter/X.
• Discord, Twitter/X

Licensing & Compatibility

• The license is not explicitly stated in the README. Compatibility for commercial use or closed-source linking is therefore undetermined.

Limitations & Caveats

• As of v0.8.0, configuration files and chat history are not automatically migrated from previous TypeScript versions, requiring users to downgrade for data access.
• macOS and Linux users need to manually install Python and Node.js environments.