Mtarsier is a free, open-source desktop application developed by Mcp360, designed to manage Model Context Protocol (MCP) servers and skills across various AI clients. Built with Rust for its backend and Tauri for the desktop framework, it features a user interface developed using React and TypeScript. This application runs locally on macOS, Windows, and Linux, providing a centralized tool for developers and AI enthusiasts to organize their AI client configurations without requiring accounts or cloud services. It simplifies interaction with numerous AI clients, including Claude Desktop, Cursor, and VS Code.

What does Mtarsier connect to?

Mtarsier is an open-source desktop application that manages configurations for Model Context Protocol (MCP) servers and skills across many AI clients. It connects to and manages settings for clients such as Claude Desktop, Cursor, VS Code, ChatGPT Desktop, Codex, GitHub Copilot, Windsurf, Antigravity, Claude Code, Copilot CLI, Gemini CLI, Codex CLI, Claude (web), and ChatGPT (web). Mtarsier itself does not implement standard MCP transport protocols but is compatible with managing configurations for clients that utilize them.

Core features and tools

Mtarsier offers several callable tools and features for managing MCP servers and skills. For server management, it includes a Unified Dashboard, Auto-Detection of AI clients, a Config Editor with JSON validation, a Marketplace for browsing servers, Auto-Backup, and Multi-Theme support. Skills management features include a Skills Manager, a Skills Marketplace, One-click Install to clients, Custom Skills creation, and Copy & Share functionality. Additionally, it provides a command-line interface tool named "tsr" for terminal-based server and skill operations.

Who is it for?

Mtarsier targets developers and AI enthusiasts who use multiple AI clients like Claude Desktop, Cursor, and VS Code. These users benefit from a unified approach to managing their MCP servers and AI skills. The application operates entirely on the user's machine, ensuring data privacy and local control over configurations. It does not collect personal data or track usage. Skip this if you only use a single AI client and manage configurations manually without difficulty.

How to get started

No credentials are required for Mtarsier's own operation, as it runs entirely locally without accounts or cloud services. It helps users set up authentication details for the MCP servers it manages, which may use methods like key-based access or OAuth 2.1.

Use cases

• Users can automatically detect installed AI clients, then visually edit, validate, and manage all MCP server and skill configurations from a single interface, removing the need for manual JSON file editing.
• Users can browse a marketplace of curated MCP servers and skills and install them into one or multiple AI clients with a single click, ensuring consistent experiences across tools.
• Users can export a ".tsr" snapshot of their MCP and skill setup, allowing teammates to import it with one click to achieve a pre-configured and consistent environment across a team.

Mtarsier provides a focused solution for managing AI client configurations and skills locally. Its open-source nature and local operation ensure user control and data privacy. The application simplifies complex configuration tasks through its unified interface and marketplace access. It is a valuable tool for individuals and teams working with multiple AI clients and MCP servers. Try managing your AI client configurations using the Unified Dashboard.