Dive AI Agent
by OpenAgentPlatform
Dive is an open-source MCP Host Desktop Application that seamlessly integrates with any LLMs supporting function calling capabilities. It allows users to extend LLMs with powerful tools and personalize AI behavior.
Last updated: 2025/3/14
What is Dive AI Agent?
Dive is an open-source desktop application that acts as an MCP (Model Context Protocol) host, enabling seamless integration between LLMs and external tools. It supports various LLMs with function calling capabilities, allowing them to access and utilize tools through MCP.
How to use Dive AI Agent?
Download the application for your operating system (Windows, MacOS, or Linux) from the latest release. Install any necessary dependencies (Python, Node.js). Configure MCP servers with tools like Fetch and Youtube-dl using the provided JSON configurations. Connect to external MCP servers via SSE if needed. Use custom instructions to personalize AI behavior.
Key features of Dive AI Agent
Universal LLM Support (ChatGPT, Anthropic, Ollama, OpenAI-compatible models)
Cross-Platform (Windows, MacOS, Linux)
Model Context Protocol (MCP) Integration
Multi-Language Support
Advanced API Management
Custom Instructions
Auto-Update Mechanism
Use cases of Dive AI Agent
Extending LLMs with web scraping (Fetch)
Enabling LLMs to download Youtube videos (Youtube-dl)
Personalizing AI behaviour with custom System prompts
Using local LLMs through Ollama
Accessing Filesystem tools for LLMs.
FAQ from Dive AI Agent
What LLMs does Dive support?
What LLMs does Dive support?
Dive supports ChatGPT, Anthropic, Ollama and OpenAI-compatible models.
What operating systems are supported?
What operating systems are supported?
Dive is available for Windows, MacOS, and Linux.
What is MCP?
What is MCP?
MCP stands for Model Context Protocol. It allows LLMs to interact with external tools and services.
How do I configure MCP servers in Dive?
How do I configure MCP servers in Dive?
You can add JSON configurations to your Dive MCP settings to enable tools such as Fetch and Youtube-dl. Examples are provided in the README.
Where can I get help or report issues?
Where can I get help or report issues?
You can join the Discord server, follow on Twitter/X, star the GitHub repository, or report issues on the Issue Tracker.