Dive - AI Developer Tools Tool

Overview

Dive is an open-source, cross-platform desktop MCP (Model Context Protocol) host and AI-agent front end that connects function-calling LLMs with tool-enabled agent workflows. It acts as a host and UI for both local and managed MCP servers (OAPHub integration), letting developers and power users run model-driven agents that invoke tools (file I/O, web fetch, shell, etc.) and switch between multiple LLM providers (ChatGPT/OpenAI-compatible, Anthropic, Ollama, and more). Dive ships in two architectures — a lightweight Tauri build (recommended) and a traditional Electron build — and provides built-in conveniences such as API key and model management, custom instructions, chat draft saving, keyboard shortcuts, and an auto-update mechanism. ([github.com](https://github.com/OpenAgentPlatform/Dive)) Designed for both end users and developers, Dive supports multiple languages and offers advanced MCP features (stdio and SSE transports), granular tool enabling, and the ability to connect to managed MCP servers via one-click OAPHub links to avoid local dependency setup. Recent releases have added built-in local tools (Fetch, File Manager, Bash), an MCP server installer agent, multi-elicitation support, and small installer improvements (for example: git is no longer a prerequisite in the latest patch). The project is actively developed and released under an MIT license. ([github.com](https://github.com/OpenAgentPlatform/Dive))

GitHub Statistics

• Stars: 1,681
• Forks: 157
• Contributors: 12
• License: MIT
• Primary Language: TypeScript
• Last Updated: 2026-01-09T01:51:47Z
• Latest Release: v0.12.5

According to the project's GitHub repository, Dive is actively maintained and has a healthy open-source footprint: ~1.7k stars, ~157 forks, and 12 listed contributors. The repo mixes TypeScript (front end/back end) and a Rust/Tauri component, with additional Python components in companion projects (dive-mcp-host) for server/host functionality. The project has an established release history (v0.12.5 released Jan 8, 2026) and a long commit history (~1,400+ commits), indicating ongoing development and iterative feature releases. Core indicators (stars, forks, releases, contributors, MIT license) and platform language breakdown are shown on the repository page. ([github.com](https://github.com/OpenAgentPlatform/Dive))

Installation

Install via npm:

git clone https://github.com/OpenAgentPlatform/Dive.git

cd Dive

npm install

npm run dev        # starts Electron development build

cargo tauri dev    # starts Tauri development build (if developing Tauri)

# For the MCP host (server) component (separate repo):

git clone https://github.com/OpenAgentPlatform/dive-mcp-host.git

cd dive-mcp-host

pip install -e .   # install Python MCP host (requires Python 3.12+)

Key Features

• Universal LLM support: connect ChatGPT/OpenAI-compatible models, Anthropic, Ollama, and others.
• Dual architecture: Tauri (small installer) and Electron builds for different use cases.
• Model Context Protocol (MCP): stdio and Server-Sent Events transport support for agent tooling.
• OAPHub integration: one-click managed MCP servers to avoid local dependency setup.
• Built-in local tools: Fetch (web requests), File Manager (read/write), and Bash executor.
• Advanced API management: multiple API keys and model switching via model_settings.json.
• Custom instructions and granular tool enable/disable for tailored agent behavior.
• Chat UX features: draft saving, keyboard shortcuts, token-usage display (recent releases).

Community

Dive maintains an active community presence: repository discussion channels (Issues, Pull Requests), an official Discord, and social links (Twitter/X, Reddit). The project shows regular releases and public changelogs; community feedback is visible on discussion threads and subreddits about LocalLLM/MCP tools where users discuss architecture changes and upgrades (for example, major overhauls and v0.8+ discussions). Contributors and maintainers accept PRs and provide build instructions for developers. For security and operations discussions (supply-chain notes, TLS options), community resources and third-party MCP directories surface practical deployment cautions. ([github.com](https://github.com/OpenAgentPlatform/Dive))

Key Information

• Category: Developer Tools
• Type: AI Developer Tools Tool