Manufact (formerly mcp-use) is a platform for deploying, testing, observing, and publishing MCP servers and MCP apps. It's built by the team behind the open-source mcp-use SDKs for TypeScript and Python and the open-source MCP Inspector. Deploys run from GitHub, the CLI, or Manufact's own MCP server, and the platform covers the lifecycle after deploy: cross-client testing, MCP-native analytics, and publishing checks for the ChatGPT Apps Store and Claude Cloud Connectors.

Under 60 seconds from git push to a live MCP endpoint. Connect the GitHub App once and every push deploys automatically, or use npx mcp-use deploy from the CLI. An agent can also drive deployments through Manufact's official MCP server.

TypeScript and Python are first-class through the mcp-use SDKs, with presets for FastMCP and other frameworks, plus a Dockerfile option for anything else. You can also connect an MCP server hosted elsewhere by URL and use testing, analytics, and publish checks without moving your hosting.

Manufact test suites exercise your deployed server's tools across clients and swap models between GPT, Claude, and Gemini, with LLM-judged pass/fail results per client and model. Suites can gate CI, so a regression in any client blocks the merge. The embedded Cloud Inspector also lets you invoke tools, browse resources, and preview widgets against real clients with no local setup.

MCP Inspector is an open-source tool maintained by Manufact for debugging MCP servers: tool testing, resource browsing, prompt testing, real-time JSON-RPC logging, and widget preview with full window.openai emulation. Run it hosted at inspector.manufact.com, locally with npx @mcp-use/inspector, or self-hosted via Docker. Inside the Manufact dashboard, the Cloud Inspector runs against your deployed server directly.

Tool calls, sessions, error rates with issue triage, and p50/p95/p99 latency per tool, resource, and prompt. Traffic breaks down by client (ChatGPT, Claude, Cursor, Windsurf, custom), client version, protocol version, and country. Session replay shows the full timeline of each session, and an internal-traffic toggle excludes dashboard and inspector calls from production numbers.

OAuth ships with six documented providers: Auth0, Better Auth, Clerk, Keycloak, Supabase, and WorkOS, plus support for any custom provider that supports dynamic client registration. Your tools receive verified user context on every call. Clients connect with OAuth 2.1 with PKCE or bearer tokens.

Manufact runs six categories of publishing checks mapped to store requirements: protocol and discovery, tool conformance, security and policy, metadata and configuration, domain/TLS/CSP, and assets. Each failed check comes with fix guidance and an autofix flow. E2E checks then run your server live inside ChatGPT and Claude to verify tool calls and widget rendering, and a generated submission pack produces listing copy, tool justifications, and reviewer test cases.

Yes. Every branch gets its own MCP URL (<slug>--br-<branch>.run.mcp-use.com/mcp), so reviewers can point a real client at the change before merge. Available on Hobby and above.

Yes. The mcp-use SDKs for TypeScript and Python (servers, clients, and agents) and the MCP Inspector are open source, with 10M+ downloads and 10k+ GitHub stars. The Manufact cloud platform builds on them.

The Free plan is $0 and includes $5/month in usage credits. Hobby is $25/month and adds preview deployments, E2E checks, and the submission pack. Startup is $250/month with all regions, and Enterprise is custom. Usage is metered ($0.10 per 1k tool-call requests, $1 per eval run, $2 per E2E check); discovery traffic like tools/list is never billed.

Yes. Connect an existing remote MCP server by URL and get the Chat interface, test suites, publish checks, and the submission pack against it. Hosting on Manufact adds deploys from GitHub, preview environments, and gateway-level analytics.