ChatGPTNextWeb/NextChat
NextChat is a lightweight, privacy-focused AI assistant platform supporting multiple AI models and operating systems, with Model Context Protocol (MCP) support for enhanced AI interactions.
Awesome MCP › Desktop & Mobile Clients
modelcontextprotocol/example-remote-client
Example Remote Client is a React and TypeScript application published by the modelcontextprotocol organisation that demonstrates what a remote Model Context Protocol client looks like in practice. It connects to several MCP servers at the same time over the HTTP and SSE transports, gathers their tools, and drives a conversational interface in which a model can call those tools inside an agent loop. The README marks the project as a work in progress and states it serves two purposes at once: a public example implementation of MCP client features that others can read and copy, and a prototyping testbed the maintainers use when trying out changes to the protocol itself. The architecture is organised around React hooks and providers, each covering one concern. InferenceProvider abstracts model inference so different backends can be plugged in, with OpenRouter as the first implementation. MCPProvider manages connections to multiple MCP servers and their lifecycle. AgentLoop implements the tool-calling and conversation flow. The remaining UI components are modular and reusable, laid out as a responsive interface with a left sidebar for servers and a main chat area. Two features are aimed squarely at developers rather than end users: real-time visualisation of tool calls as they happen, and MCP debugging with message tracing so the raw protocol traffic can be inspected while a conversation runs. The stack is React 18 with TypeScript, Vite for building, Tailwind CSS for styling, the official MCP TypeScript SDK for the protocol layer and Vitest for tests, with scripts for dev server, production build, preview, lint and test including a test UI. It needs Node 18 or higher and works with npm, yarn or pnpm. It is most useful to anyone writing their own MCP client.
https://github.com/modelcontextprotocol/example-remote-client
Also in Desktop & Mobile Clients
CherryHQ/cherry-studio
Cherry Studio is a cross-platform desktop client supporting multiple LLM providers and featuring an MCP server, AI assistants, document processing, and practical AI tools for enhanced user experience.
agentscope-ai/QwenPaw
Self-hosted personal AI assistant in Python with a desktop app, terminal UI and chat-channel connectors, whose driver layer works as an MCP host that routes agent tool calls to external MCP, A2A and ACP servers.
nukeop/nuclear
Nuclear is a free, open-source music player for desktop that allows users to search, stream, and manage music from various sources, featuring a powerful plugin system and an MCP server for AI agent...
electerm/electerm
Electerm is a versatile open-source multi-platform client for terminal, SSH, SFTP, FTP, Telnet, serial port, RDP, and VNC connections with integrated AI assistant support via the Model Context Protocol (MCP) widget.
rikkahub/rikkahub
RikkaHub is a native Android chat client app supporting multiple large language model providers with Model Context Protocol support, multimodal input, and advanced customization features.
punkpeye/awesome-mcp-clients
A curated collection of production-ready and experimental Model Context Protocol (MCP) clients enabling secure AI model interactions with local and remote resources through standardized implementations.
ThinkInAIXYZ/deepchat
DeepChat is an open-source AI chat platform that unifies multiple large language models with advanced search enhancement and Model Context Protocol (MCP) support for tool integration and code execution.