mTarsier - Open-source platform for managing MCP servers and clients
byβ’
Free, open-source desktop app that auto-detects every AI client on your machine β Claude Desktop, Cursor, Windsurf, VS Code and more. Manage all your MCP server configs in one place, install from the marketplace, and back up with one click. macOS, Windows & Linux.



Replies
YourGPT
Hey Product Hunt! π
Today we're launching mTarsier, the open source MCP manager we wish existed when we started building with MCP.
The problem we kept hitting: Every AI client β Claude Desktop, Cursor, Windsurf, VS Code β has its own config file. Adding, removing, or enabling an MCP server means hunting down JSON files, switching between apps, and praying nothing breaks. It's chaos, and it kills momentum.
The problem isn't new β the community has been screaming about it:
Garry Tan (YC President): "MCP sucks honestly. Toggling it on and off, the auth sucks." A dev on X: "why does every AI client handle MCP config differently?? I just want ONE place." Hacker News has a 145-point thread: "MCP is a fad."
But here's the other side:
OpenAI, Google DeepMind, AWS, Microsoft, and Cloudflare all backed MCP. Downloads went from 100K to 8 million in 6 months. Fortune 500s are deploying it. The Linux Foundation now governs it. Jensen Huang called it something that "completely revolutionized the AI landscape."
MCP isn't a fad. The protocol won. The tooling just hasn't caught up β yet. That's what mTarsier is here to fix.
What mTarsier does: One platform to manage your entire MCP ecosystem β across every client, from one place.
π Auto-detects Claude Desktop, Cursor, Windsurf & VS Code on install
π₯οΈ Visual dashboard β see all your MCP servers and their health at a glance
ποΈ Built-in MCP Marketplace β browse & install MCPs without touching a config file
π Multi-client sync β manage all your AI clients together
β‘ One-click install & enable β no manual JSON editing, ever
π¦ Team sharing β export your entire MCP workflow as a .tsr file and share it instantly
π€ Agent-native CLI β install tsr and let Claude or your AI agent manage its own MCPs directly
π₯οΈ Cross-platform β macOS, Windows & Linux
π Fully open source & free β forever
Built on Rust. Lightweight, fast, and native.
MCP unlocks your AI agents. mTarsier unlocks MCP.
We'd love your feedback β what MCP clients or features should we prioritize next? Drop a comment below π
Auto-detecting every AI client on the machine β Claude Desktop, Cursor, Windsurf, VS Code β and unifying all their MCP config files into a single visual dashboard eliminates the most painful part of the MCP ecosystem right now, which is manually editing scattered JSON configs and hoping nothing breaks. The .tsr file format for exporting and sharing entire MCP workflows across teams is a smart standardization move; with the agent-native CLI that lets AI agents manage their own MCPs directly, how do you handle security boundaries β can an agent install any MCP server from the marketplace autonomously, or does it require human approval for new tool installations?
YourGPT
mTarsier : A free and open-source platform built for the community making it much easier for developers to build and manage MCP setups without the usual hassle. Really useful and thoughtfully done.
Feeling proud to be part of a team that truly cares about the community.
YourGPT
@neha_8Β Thanks for your comment!
MCP tooling is still pretty fragmented β a unified manager for both servers and clients is pretty cool. Does it support auth config per-provider, or is that still on the roadmap?
YourGPT
@abhinavrameshΒ The feature is available on our v1 Release. Auth config can be configured individually for each MCP server.
Trufflow
How does multi-client sync work exactly? Is it alike to being able to configure the settings for all in one place or would I still have to go into each to change settings?
Banyan AI Lite
How do you manage trust, security, and compatibility across such a broad ecosystem of tools in MCP360? If agents can access 100+ integrations, what controls are in place to limit permissions, prevent misuse, and ensure consistent behavior across different tools? Also, how do you handle versioning and reliability so workflows donβt break when underlying tools change?
Anyway, upvote from me + good luck
Curiousβcan mTarsier handle team-level permissions for shared MCP workflows, or is it per user for now?
The .tsr export is the feature I didn't know I needed. I've been manually copying MCP configs between my work laptop and a home server, and it's exactly the kind of thing that breaks silently when you forget one entry.
Quick question on multi-client sync: if I have Cursor on one machine and Claude Desktop on another, does the sync work across machines, or is it local to a single device for now?
@rohitjoshi Really interesting direction.
What stands out is that this feels like more than just a configuration layer for MCP.
If MCP becomes the standard interface for how agents access tools, then a system that manages MCP across clients starts to look less like a convenience layer and more like a control plane over what capabilities agents can actually use.
That feels like a deeper shift.
Instead of just simplifying fragmented configs, you're effectively centralizing how tool access, environments, and workflows are defined across agents.
Curious how you think about this long term.
Do you see mTarsier staying primarily as a developer tool for MCP management, or evolving into a more foundational control layer for agent capabilities and environments?
Does it auto detect new MCP clients when they get installed or do you need to manually add them? Congrats on the launch!