mTarsier

What is mTarsier?

mTarsier is an open-source platform for managing MCP servers and clients on your computer, helping multiple AI tools use the correct MCP connections without editing scattered configuration files. It focuses on the practical problem of managing where MCP server settings live across different clients.

The core purpose of mTarsier is to provide a unified view and editor for MCP configuration, detect supported AI clients automatically, validate config changes, and make it easier to install MCP servers through a marketplace-style flow.

Key Features

• Auto-detection of AI clients: Scans your machine to find supported clients (e.g., Claude Desktop, Cursor, Windsurf, VS Code) without manual setup.
• Unified dashboard across clients: Shows MCP servers installed across multiple AI clients in a single interface, so you can see what’s active and where it’s configured.
• Config editor with validation: Read/write MCP config files with syntax highlighting and live JSON validation to help catch errors before they break a tool.
• Marketplace for MCP servers: Browse curated MCP servers and install them into a chosen client with a single click instead of copying JSON snippets between files.
• Rollback support via backups: Creates an auto-backup before changes and supports rolling back in one click if a configuration causes issues.
• CLI tool (tsr): Includes a terminal workflow to list servers, inspect clients, install from the marketplace, and edit configs using the tsr command.

How to Use mTarsier

1. Download and install mTarsier on macOS (Apple Silicon and Intel), Windows, or Linux.
2. Open the app to run client detection: mTarsier scans for installed AI clients and presents them in its dashboard.
3. Manage MCP configs in the unified editor: Use the built-in config editor to view or edit MCP server settings with live JSON validation.
4. Install MCP servers via the marketplace: Choose an MCP server and install it into the selected client.
5. (Optional) Use the CLI: Install the tsr CLI tool from mTarsier’s settings (Settings → CLI Tool → Install tsr) for terminal-based management.

Use Cases

• Stop breaking tools due to scattered JSON: Instead of editing raw JSON files across multiple locations, use mTarsier’s unified editor and validation to reduce the chance of a typo breaking multiple clients.
• Set up MCP servers for several desktop clients: Configure MCP servers once per client using a single dashboard view, rather than manually hunting down each client’s configuration path.
• Install a browsing or memory MCP server across clients: Use the marketplace flow to install an MCP server into the specific client you want to use (e.g., different tools may be used for different workflows).
• Diagnose what’s installed and where: Quickly determine which MCP servers are active across clients and identify the client they belong to using the dashboard.
• Share a working setup with teammates: Export a .tsr snapshot so teammates can import it in one click and pre-configure clients.

FAQ

• Is mTarsier really free? The page states that mTarsier is free and open-source.

• Does mTarsier require an account or send configuration data to the cloud? The product page says it runs locally and requires no account. It also states the software is free and open-source, but it does not provide detailed data-transmission behavior in the excerpt shown.

• Which operating systems are supported? The page lists macOS (Apple Silicon and Intel), Windows, and Linux.

• What does mTarsier detect? It automatically detects supported MCP clients such as Claude Desktop, Cursor, Windsurf, VS Code, and others (the page notes “and more” and also lists additional web and desktop variants).

• How do I use the terminal tool? mTarsier includes a CLI tool called tsr. The page indicates you can install it from the app via Settings → CLI Tool → Install tsr CLI.

Alternatives

• Manual editing of MCP config files: Directly edit each client’s MCP JSON configuration in their respective folders. This is flexible but typically involves scattered paths and higher risk of breaking a client due to invalid JSON.
• Generic JSON editors and validators: Use an external editor/linter to validate JSON and make changes, but you still must locate and update each client’s configuration file separately.
• Client-specific MCP support (within each AI client): Some tools may include their own MCP configuration mechanisms. This keeps setup within one application but usually doesn’t provide a cross-client management view.
• Terminal-only MCP management (custom scripts): Use command-line scripts to automate config edits. This can work for power users but requires more setup effort and maintenance than a unified manager.