
Golf
The enterprise firewall for MCP providers
524 followers
The enterprise firewall for MCP providers
524 followers
Golf is the enterprise control plane for MCP. It gives security and IT teams full visibility into how AI connects to enterprise systems — with policy enforcement, real-time threat blocking, and a complete audit trail. Discover, enforce, audit. End-to-end.









Tech To The Rescue
Prompt injection is the new SQL injection and most teams don’t even realize they’re already exposed.
Golf is solving a very real pain at the core of MCP adoption: security and trust.
Love the clarity of the problem statement and how deep you’ve gone into the “poisoned prompt” vector.
Congrats Wojtek & Antoni, feels like you’re building the missing firewall of the agentic internet 👏
Golf
@tomik99 Couldn’t agree more! Prompt injection really is the new SQL injection. Most teams don’t realize how exposed their MCP servers are until we show them. That’s exactly why we built Golf. Appreciate the kind words!
Golf
@tomik99 thanks Tomek!
cubic
Congrats on the launch!
Quick question: does Golf work with any MCP server out of the box, or does it require specific integration work to get up and running?
Golf
@paul_sangle_ferriere1 Thanks Paul! Yep - Golf works with any MCP server right out of the box. No code changes needed. You just point your MCP traffic through Golf, and we start inspecting and securing all the content automatically.
Golf
@paul_sangle_ferriere1 thanks Paul! Cubic is great btw:)
DiffSense
Congratulations on the launch, team! 🎉
Just curious, how does Golf Firewall work? Is it an SDK that needs to be installed on the server side, or does it act as a proxy for client-server requests and responses?
Also, would it work with custom-built clients too, beyond tools like Cursor or Claude?
Very timely launch, industry is in a dire need of such solutions. Kudos!
Golf
@prastik Thanks, Prastik!
Golf sits transparently in front of your MCP server. It’s a lightweight layer that inspects and filters all inbound and outbound traffic. You deploy it as a separate instance (often on-prem, in the same environment as your MCP servers), and it doesn’t require any SDK or code changes. It works seamlessly with any MCP server and any MCP client, even custom-built ones.
Golf
@prastik thanks Prastik!
Congrats on the launch guys!
I had a couple of questions and thoughts. Since you’ve been building in the MCP space since February, what patterns have you noticed so far? From the outside, it feels like MCPs still haven’t fully taken off. With Claude Code (and others) getting increasingly good at tool use, do you think the landscape is shifting in a way that might eventually surpass MCPs altogether?
There was also Google’s agent2agent release, but I haven’t seen much discussion about it since. And looking beyond the MCP ecosystem, I’m curious whether your solution could also function as a lower-level “system firewall” between computer-use agents and the tools they interact with.
Stormy
MCP security is obvious issue no one approached yet. Glad Golf tackled it
Golf
@karmedge IT IS! Thanks for your support and upvote!
Golf
@karmedge totally agree!
Congrats 🎉 This looks like a game-changer for MCP security—love seeing teams tackle real-world risks like prompt injections.
Webflow SaaS Showcase
Love the positioning - security for MCP servers is such a timely need. Congrats on the launch! 🚀