Connecting Threaded MCP

AI tools are only as useful as the data they can reach. A general chat tool can talk about assembly processes in the abstract, but it cannot look at yours — and it certainly cannot update them. Bridging that gap by pasting exports into a chat window is slow and goes stale the moment you do it.

Threaded MCP is the connection that closes the gap. It lets AI tools work with your live Threaded data — work instructions, the Map, parts and tools, issues, reports — acting as you, with exactly the permissions you already have in the app. It is what makes the in-app AI Assistant able to change things rather than only advise, and it is how tools like Cursor and Claude can read your process directly.

#Key Concepts

Threaded MCP — Threaded’s hosted service that exposes your organization’s data and workflows to AI tools through the Model Context Protocol (MCP), an open standard for connecting AI tools to outside systems. It always acts as the signed-in user.

Hosted server — A remote endpoint Threaded runs at mcp.threadedmfg.com. There is nothing to install or run yourself.

Sign-in — The first time a tool connects, a browser window opens so you can sign in with your Threaded account. The tool keeps a secure session until you disconnect or it expires.

Per-organization — Connections apply to the organization you made them in. They do not carry over to other organizations you belong to.

#Inside Threaded

For most organizations there is nothing to do: Threaded MCP connects automatically using the login you already have, so the AI Assistant can act on your process from the first message.

You can see the state of the connection from the icon at the top of the AI Assistant sidebar, which reads Threaded MCP connected, disconnected, or expired. Clicking it opens the connections page. If the assistant needs the connection and it isn’t live, it will show you a Connect card in the conversation — one click and it picks up where it left off.

To manage connections directly, open Admin → MCP Connections (available to editors and owners). Each connector shows a status chip — Connected, Expired, or Disconnected — with a Connect, Reconnect, or Disconnect button. If you disconnect on purpose, Threaded respects that and will not silently reconnect you.

For an overview of what the assistant does with the connection, see The AI Assistant.

#Connecting from Cursor

Cursor supports remote MCP servers. Add Threaded once, and Cursor handles sign-in the first time you use it.

1. Open Cursor Settings → MCP and add a new MCP server.
2. Set the server URL to https://mcp.threadedmfg.com/mcp.
3. Save. The first time Cursor talks to the server, a browser window opens for you to sign in with your Threaded account.
4. Confirm the server shows as connected. Cursor’s MCP Logs will tell you what went wrong if it doesn’t.

You can also define the server in a config file — .cursor/mcp.json in a project, or ~/.cursor/mcp.json for all projects:

{
  "mcpServers": {
    "threaded": {
      "url": "https://mcp.threadedmfg.com/mcp"
    }
  }
}

Restart Cursor after editing the file. The first tool call triggers sign-in.

#Connecting from Claude

Claude supports remote MCP servers with the same URL.

• Claude Desktop — open Settings → Connectors, add a custom remote server with the URL above, and complete the Threaded sign-in when the browser opens. Verify the connection is active before asking Claude about your data.
• Claude Code — use claude mcp add with HTTP transport, or follow Claude’s current remote MCP setup flow for the same URL.

Menu names vary between versions; the server URL does not.

#What Connected Tools Can Do

Once connected, a tool can work within your organization and your permissions:

• Read your process — work instructions, Map nodes, parts, tools, issues, reports, and the datasets reports read from.
• Run Threaded operations — the same structured commands the Threaded CLI exposes, executed as your user.
• Check the connection — confirm which environment it is connected to and whether the session is still valid.

If you want a connection that can only read, point the tool at the read-only endpoint below instead of /mcp. Inside Threaded, changes the AI Assistant proposes still go through the Execute confirmation regardless of which endpoint is in use.

#Server URLs

Purpose	URL
Read and write	https://mcp.threadedmfg.com/mcp
Read only	https://mcp.threadedmfg.com/mcp/readonly

Use the URL that matches the Threaded app you sign into — app.threadedmfg.com for production. Threaded staff and preview environments use a different host.

Visiting https://mcp.threadedmfg.com in a browser gives you a short overview of the service and its endpoints.

#Troubleshooting

• Connection failed or expired — disconnect and connect again from Admin → MCP Connections, or remove and re-add the server in Cursor or Claude.
• Wrong organization or account — the connection runs as the Threaded user who signed in, and applies only to the organization it was made in. Sign out and reconnect to change either.
• Your tool never prompts for login — check that the URL ends in /mcp (or /mcp/readonly) and that the tool supports remote MCP servers.
• Still stuck — contact [email protected] with the tool you were using and roughly when the failure happened.

#Why Connect Threaded MCP?

Inside Threaded, this connection is what turns the AI Assistant from an advisor into a collaborator that can actually update your process. Outside Threaded, it brings your live process data into the tools you already use for analysis and automation, without exporting spreadsheets or copying context by hand. Either way, access stays scoped to your user and your organization, and you can disconnect whenever you want.