TL;DR

Manufact is the MCP cloud for dev teams to ship MCP Apps and Servers in production. Connect a GitHub repo and Manufact handles deployment, security, observability, and scaling, so teams can ship in minutes instead of weeks.

One codebase, deployed everywhere users and agents already live: the ChatGPT Apps Store, Claude Connectors, Gemini, Cursor, Claude Code, and more.

We're also the creators of mcp-use SDK, the most popular high-level MCP framework for building MCP Apps and Servers.
(mcp-use SDK is to the Model Context Protocol SDKs what Next.js is to React!)

It’s open-source and used by developers at 20% of US500 and companies like 6sense, Elastic, IBM, Innovacer, Intuit, NVIDIA, Oracle, Red Hat, Tavily, and Verizon are building with us.

The ask

If you're building MCP Apps or Servers, or you know someone who is, we'd love to talk.

https://cal.com/team/manufact/chat
or shoot a message to us: +1 (628) 899-2498
or founders@manufact.com

Our story

When MCP came out, we couldn't believe such a powerful protocol was locked into IDEs and Claude Desktop.

So we built mcp-use, the open source framework that lets any developer build MCP Apps and Servers in code, in a structured and composable way.

It got adopted fast: 7M+ downloads across Python and TypeScript, almost 10k GitHub stars, and dev teams at top companies shipping with it.

But OSS is only half the story. Once teams had something working, they hit the wall every infra team knows: deployment, auth, secrets, observability, scaling. Stitching that together for MCP is a multi-week project. That's what Manufact solves.

Make something agents want

MCP unlocks something most products have never had access to before: real estate inside ChatGPT and Claude.

800M+ users already spend their day in those chats.
With MCP, a product can show up there as a native app or connector, with no separate UI to build, no app to install, no new login flow.
The user just asks, and your product does the work.

Problem

The catch is that "build once, ship everywhere" is its own infra problem. ChatGPT Apps, Claude Connectors, Gemini, and the long tail of agent surfaces (Cursor, Claude Code, Codex, OpenAI Agents, Mastra, LangChain, Vercel AI SDK) each have their own quirks, auth flows, and submission requirements. Manufact lets you build one codebase and ship it to all of them.

What Manufact does

Connect your MCP repo once. Every push auto-deploys.

1. Build: Scaffold a full MCP App or Server with npx create-mcp-use-app or drop in existing MCP code unchanged.
2. Deploy: GitHub App connects once. Every push auto-deploys to Manufact Cloud. Preview URL per branch. Custom domains with SSL handled.
3. Publish. Get your app in front of users in the ChatGPT Apps Store and Claude Connectors. Submission assets, marketplace checklists, and an embeddable chat widget for your own site, docs, or campaign pages.
4. Iterate. Cloud Inspector debugs from any browser against real clients. Swap models. Run evals on every deploy.
5. Monitor. Analytics, session replay, JSON-RPC traces, and alerts on regressions, so you catch issues before users do. Less than 60 seconds from git push to a live MCP App or Server, available across ChatGPT, Claude, Gemini, and any agent surface that speaks MCP.

Ask

Do you know anyone already using MCP servers internally or building products powered by AI agents?

For the first few companies we will personally create/integrate MCP servers you need, have everything running through our platform, either on cloud or your own infrastructure. We even help you integrate our SDK into your code.

What do you think? Can we help you?

Book a call: https://cal.com/team/manufact/chat

Shoot a message to us: +1 (628) 899-2498 or founders@manufact.com