Overview

OneIQ's MCP (Model Context Protocol) server lets your AI application query OneIQ data directly. To connect, your local AI application needs to add the OneIQ MCP Server using a client ID that OneIQ provides you.

Before OneIQ can issue a client ID, we require the redirect URI(s) your application uses for OAuth sign-in. Since this may vary from application to application, setup involves two parts:

Providing OneIQ with your application's redirect URI(s)
OneIQ generates a client ID that you can use to complete the setup and connection to the OneIQ MCP Server

The exact screens vary a little between applications, but the overall process is the same everywhere: find your application's redirect URI and send it to OneIQ Support, wait for OneIQ to provide a client ID, then point your application at the OneIQ MCP server address, enter that client ID, and sign in with your OneIQ account when prompted.

What you'll need

The redirect URI(s) your local AI application uses for OAuth sign-in — you may need to refer to the documentation of your AI application (if available) to find the exact redirect (callback) URI(s). Alternatively, you can try connecting your AI application to the OneIQ MCP server as if dynamic client registration were supported. The attempt will fail, but depending on your application, the resulting error may reveal the redirect URI(s) it expects; if not, OneIQ support can look up the logs for the failed attempt to determine the redirect URI(s) your application provided.
The OneIQ MCP server address — https://mcp.oneiq.com
The client ID OneIQ has assigned for your application — a long string like f0c077f8-79b1-46ec-be13-ccdd51a8d70b. You won't need a client secret; leave that field blank if your application asks for one.
Your OneIQ account credentials, which you'll use to sign in and approve access when prompted.
An AI application that supports connecting to remote MCP servers over OAuth — for example LM Studio, VSCode - GitHub Copilot, or another MCP-compatible client.

General setup steps

Part 1: Provide your redirect URI to OneIQ

Refer to the documentation of your AI application (if available) to find the exact redirect (callback) URI(s) it uses for OAuth sign-in.
Alternatively, you can try connecting your AI application to the OneIQ MCP server as if dynamic client registration were supported — the attempt will fail, but depending on your application, the resulting error may reveal the redirect URI(s) it expects.
Send the redirect URI(s) to OneIQ Support, along with the name of the application you're connecting. If you weren't able to determine the redirect URI(s), please let OneIQ Support know about the failed connection attempt so that they can determine the redirect URI(s) your application provided.
Wait for OneIQ Support to confirm they've registered a client ID for you and send it to you — you'll need it for Part 2 below.

See Appendix A at the end of this article for notes on where specific applications (LM Studio, Microsoft Foundry, GitHub Copilot for VS Code, Open WebUI) show their redirect URI.

Part 2: Connect using your client ID

Add a new server connection (or return to the one you started in Part 1). Give it a name (e.g. “OneIQ-MCP”) and enter https://mcp.oneiq.com for the server address.
Look for an “Advanced,” “Authentication,” or “Custom OAuth client” option — this is usually a toggle or expandable section, since most applications hide it by default.
Enter the client ID provided by OneIQ in the field usually labeled Client ID or CLIENT_ID. Leave the client secret field empty — OneIQ's client registrations don't use a secret.
Save the connection. Your application should then prompt you to sign in — this typically opens a browser window.
Sign in with your OneIQ account credentials, review the permissions being requested, and approve access.
Once approved, you'll be returned to your application, which should now show the connection as active and ready to use.

For pointers specific to LM Studio, Microsoft Foundry, GitHub Copilot for VS Code, and other applications, see Appendix A: Application-Specific Notes at the end of this document.

Troubleshooting

If sign-in opens a browser window that shows an error, try again from your application — a fresh attempt often resolves transient issues.
Double check that the client ID was entered exactly as provided, with no extra spaces, and that the client secret field is left empty.
Make sure your browser isn't blocking pop-ups for the sign-in step.
If you continue to see an error after a couple of attempts, please reach out to OneIQ Support with a screenshot of the error message — this helps us diagnose the issue quickly.

If you have any questions during setup, please contact OneIQ support.

Appendix A: Application-Specific Notes

The following examples show how to connect to the OneIQ MCP server using a client ID across a range of applications.

Claude Code

Claude Code connects to remote MCP servers from the command line using its claude mcp commands. Signing in requires a real interactive browser session, so part of this process runs in a plain terminal window rather than inside an active Claude Code session — Claude Code will tell you to do this if you try to authenticate from within a chat.

Add the OneIQ MCP server with claude mcp add from a terminal, giving it a short name (e.g. oneiq), the server address, and the client ID OneIQ provided you:

claude mcp add --transport http --client-id <client ID> mcp-oneiq https://mcp.oneiq.com
Run claude mcp list to confirm the server was added — it will show as configured but Needs authentication.
Open a plain terminal window (not a Claude Code chat), cd into the same project directory you ran claude mcp add from, and run:

claude mcp login mcp-oneiq

If you run claude mcp login from inside an active Claude Code session, it will fail — that session's shell has no TTY to receive the OAuth callback, so this step must run in a normal terminal window. Also if you are not in the project directory when running the above command, Claude will return a "No MCP server named "mcp-oneiq..." error.
This opens your browser to OneIQ's sign-in page (or prints a URL to open manually if it doesn't open automatically). Sign in with your OneIQ account credentials and approve access and when prompted, click on Allow to Authorize Claude Code
The terminal should confirm Authenticated with 'oneiq'. Its tools are now available in Claude Code.
Back in your Claude Code session, run claude mcp list again (or just ask Claude to use the tool) to confirm the connection now shows ✓ Connected.

LM Studio (Bionic interface)

LM Studio's newer “Bionic” agent interface has a dedicated settings screen for this:

Go to Settings → Connected Apps, then add a new MCP server.
Enter the server address as the Server address.
Toggle on Show advanced options to reveal the OAuth client details section, where you'll enter the client ID.
Leave Client secret blank.

LM Studio Bionic's MCP server configuration screen, showing the Server address and OAuth client details fields.

If your settings don't look like the screen above, you're likely on LM Studio's classic (non-Bionic) interface, which configures MCP servers differently — through an mcp.json file rather than a settings screen. See below.

LM Studio (classic interface)

Click + Install and select Edit mcp.json. A warning popup may appear about adding MCP servers from an untrusted source — click Got it.

Add an entry for OneIQ with the server address and your client ID, following this pattern:

{
  "mcpServers": {
    "oneiq": {
      "url": "https://mcp.oneiq.com",
      "auth": {
        "CLIENT_ID": "<Enter the client ID provided by OneIQ here>"
      }
    }
  }
}

Save the file (Ctrl+S). LM Studio will note that saving reloads your MCP servers.
Once saved, find the new entry (e.g. mcp/oneiq) under Integrations in the right-hand sidebar and make sure it's toggled on.
You'll see an “Authentication Required” prompt. Click Authenticate (or, if a dialog appears, Trust and Open in Browser) to open your browser and sign in to OneIQ with your account credentials to complete setup.

Editing mcp.json (top) and the sign-in prompt that follows after saving (bottom), in LM Studio's classic interface.

GitHub Copilot for VS Code

This refers specifically to GitHub Copilot's MCP support within Visual Studio Code (not other Copilot integrations, such as Visual Studio or JetBrains IDEs).

In the GitHub Copilot chat window, click Open Customizations (the gear icon).
Click MCP Servers, then click the + button on the right.
Choose “HTTP (HTTP or Server-Sent Events)” — connect to a remote HTTP server that implements the MCP protocol.

Choosing the “HTTP (HTTP or Server-Sent Events)” MCP server type in VS Code.
Enter https://mcp.oneiq.com for the Server URL and press Enter.
Accept the pre-generated Server ID, or enter your own.
A prompt may appear stating “Dynamic Client Registration not supported.” Click Copy URIs & Proceed — this copies the redirect URIs VS Code will use, which is what OneIQ needs to register a client ID for you (see Part 1 of the General setup steps). If you don't already have a client ID from OneIQ, send these copied URIs to OneIQ support now and wait for them to reply with one before continuing.
In the Add Client Registration Details box, enter the client ID provided by OneIQ and press Enter. Press Enter again when prompted for a client secret — this can be left blank.

Entering the client ID OneIQ provided in the Add Client Registration Details prompt.
A prompt will appear asking you to authenticate to the OneIQ MCP server. Click Allow, then click Open to connect to OneIQ.
Sign in to OneIQ with your existing account credentials to complete the process.

You may need to click Start Server to enable the MCP server connection in VS Code.

Open WebUI

These steps apply to the standard Open WebUI server (run via Docker or pip) with native MCP support. Open WebUI Desktop does not yet support MCP connections as of this writing — see the note at the end of this section. If you're running Open WebUI in Docker, make sure the WEBUI_SECRET_KEY environment variable is set before you begin — without it, OAuth tokens can fail to decrypt after a restart, requiring you to re-authenticate.

Make sure you're running Open WebUI version 0.6.31 or later — native MCP support requires this version.
Click on your profile Avatar and go to Admin Settings → Integrations
Click the + (Add Server) button.
Set the Type dropdown to MCP (Streamable HTTP) — not “OpenAPI.”
Enter https://mcp.oneiq.com for the Server URL.
For Auth, choose OAuth 2.1 (Static)
Enter the Client ID provided by OneIQ. Leave Client Secret blank.
If there's an OAuth Server URL field, enter https://id.oneiq.com
Click Register Client, then Save. Restart Open WebUI if prompted.
As a regular user, go to + → Integrations → Tools in the chat window and enable the tool there.

If you're running Open WebUI Desktop, note that as of the current release (v0.0.20), the Integrations screen only offers External Tool Servers (OpenAPI-compatible) and Terminal connections — there is no MCP (Streamable HTTP) option yet. Connecting to OneIQ's MCP server currently requires the standard Open WebUI server (via Docker or pip) rather than the Desktop app. Check for Desktop app updates, as MCP support may be added in a future release.

Also note: OAuth-secured tools can't be set as a default tool in Open WebUI, since the sign-in flow needs to happen interactively in the browser.

Other applications

If your application isn't listed here, check its documentation for “MCP server” or “remote tool” configuration. Any application that supports the MCP standard should offer a similar way to add a server address and, for servers requiring it, a custom OAuth client ID.

Connecting Your Local AI Application to OneIQ