Dive by Hyundaicard Co.ltd. is a free MCP Host Desktop Application designed for integrating Large Language Models with system functionalities. It operates as an AI Agent desktop application, providing tools for web requests, file management, and command execution. Developed for multiple platforms, Dive supports the Model Context Protocol, enabling connections with various AI clients. This application targets developers and citizen developers seeking to build and deploy intelligent agents on their local systems.

What does Dive connect to?

Dive by Hyundaicard Co.ltd. connects to Large Language Models (LLMs) that support function calling capabilities, acting as an MCP Host Desktop Application. It supports the Model Context Protocol (MCP) using both stdio and SSE modes, built on JSON-RPC 2.0 with Streamable HTTP transport. Compatible AI clients include ChatGPT, Anthropic, Ollama, and OpenAI-compatible models. Other MCP-capable AI assistants like Claude and Cursor are also compatible with MCP servers.

Core features and tools

Dive provides essential callable tools out-of-the-box for AI agents. These include Fetch for web requests, File Manager for reading and writing files, and Bash for command execution. Users can also add, edit, or delete custom tools directly within the MCP Server. The application supports various authentication methods for LLM providers, including API keys, and offers OAP Cloud Integration via OAPHub.ai for managed MCP servers.

Who is it for?

Dive is intended for developers, citizen developers, and users who aim to integrate LLMs with various system tools to create and deploy intelligent agents. It provides a unified desktop application experience across Windows, macOS, and Linux, supporting cross-platform AI agent development. This application suits those looking to extend LLM capabilities beyond their native environments. Skip this if you are not interested in integrating LLMs with local system functions or managing custom AI agent tools.

How to get started

Access is granted through API keys for LLM providers. For local setups, configuration of LLM API keys and MCP servers can be done via the UI or by editing JSON configuration files.

Use cases

• Integrating LLMs with local system functionalities like web requests using the Fetch tool, file management through the File Manager, and command execution with Bash.
• Customizing AI agent capabilities by allowing users to add, edit, or delete their own tools directly within the MCP Server.
• Facilitating cross-platform AI agent development by providing a unified desktop application that runs on Windows, macOS, and Linux environments.

Dive by Hyundaicard Co.ltd. offers a desktop solution for connecting LLMs to local system functions. It enables developers to extend AI agent capabilities through callable tools and custom additions, supporting various AI clients and operating systems. The application provides a structured approach for integrating advanced language models with practical system interactions.