Manufact
Cloud Articles
SSO, Two-Factor Sign-In and Passkeys in Manufact Cloud
Every Manufact Cloud account can now require a second step at sign-in, with an authenticator app, an emailed code, backup codes or a passkey. Enterprise teams can sign in through their own OIDC or SAML identity provider and make it the only way in for their domain.
How to Get AI Agents to Send Feedback on Your MCP Server
An agent can recover from a tool failure and still have something worth reporting. Collect that feedback in the Manufact dashboard with send_feedback.
Introducing Manufact MCP v2.0
Manufact MCP v1 made deployment decisions the agent could not see. V2.0 splits them into small tools that report facts, and every error names the next step.
How Do You Measure MCP Server Quality?
A free MCP quality score checks your server's tools, protocol, metadata, and discoverability. See what Manufact measures, what the score misses, and how to improve it.
Test Your MCP Server in Claude Code, Codex, and ChatGPT
Manufact test suites now run your MCP server inside Claude Code, Codex CLI, and Pi, next to ChatGPT, Claude, and the Inspector. Every run is scored by an LLM judge that tells you what to fix.
MoChiPi: Connect Your MCP Servers Once
The easiest way to take your MCPs across clients.
The 8 Best MCP Deployment Platforms in 2026
Manufact, Cloudflare Workers, Vercel, Smithery, MintMCP, Stainless, Zapier MCP, and MCPHosting.io compared on languages, testing, analytics, auth, publishing checks, and developer workflow.
MCP Deployment Platform Benchmark: 8 Platforms, 12 Metrics
Eight MCP platforms scored on 12 measurable metrics with a published rubric and a source link for every value. Manufact 95, Cloudflare 50, Smithery 44, MintMCP 42, Vercel 33.
Build vs. buy: how to run MCP infrastructure yourself
An MCP server is easy to stand up and hard to operate. These are the five layers you take on when you run MCP in production yourself, and what a web app would get for free.
We Made MCP Testing Less Painful
Testing MCP servers is painful: slow install loops, behavior that differs across clients, and no good way to know if your agents are calling tools correctly. Here's how we fixed it.