MCP Manager

Centralize your AI tools. MCP Manager is a lightweight desktop app to manage, sync, and backup MCP server configurations across your AI tools like Gemini CLI, Claude Code, etc in one unified dashboard

Comment

Inspiration

The Model Context Protocol (MCP) is revolutionary, but its implementation is currently fragmented. Developers using multiple AI tools—like Claude Code, Cursor, and Windsurf—face a "Configuration Tax." Every tool stores its config in a different hidden directory, requiring manual copying and fragile edits. I built MCP Manager to eliminate this friction, providing a "Control Plane" for the MCP ecosystem that makes powerful AI tools accessible with a single click.

How I built it

The project is architected for speed and a tiny footprint:

  • Backend: Powered by Rust and Tauri v2. Rust ensures memory-safe filesystem operations for sensitive config files, while Tauri keeps the total binary size under 10MB.
  • Frontend: Built with Vanilla JS and a custom CSS Glassmorphism design system to provide a premium, "AI-native" look and feel without the bloat of heavy frameworks.
  • Logic: I implemented a robust abstraction layer that detects tool installations across the macOS filesystem and parses both JSON and TOML configurations dynamically.

Challenges I faced

A major challenge was the Path Discovery Problem. Each editor uses obscure paths like: $$\text{Path}{\text{VSCode}} = \text{Library/Application Support/Code/User/mcp.json}$$ Mapping these across 8+ tools required deep OS-level research. Another challenge was Conflict Resolution: merging duplicate servers in the "All Tools" view while ensuring a toggle on a grouped card accurately propagated status changes to $n$ underlying files simultaneously: $$\text{Action}(S) \rightarrow \sum{i=1}^{n} \text{Update}(\text{Tool}_i, \text{Config}_S)$$

What I learned

I gained deep experience in Tauri’s IPC (Inter-Process Communication) and systems programming in Rust. More importantly, I learned that developer experience (DX) is often about solving the "boring" problems—like config management—so that creators can focus on the exciting parts of AI.

Built With

Share this project:

Updates