Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.scaling.cloud/llms.txt

Use this file to discover all available pages before exploring further.

OpenAI is rolling out native MCP Connector support in ChatGPT. Where Connectors are not available, you can wrap the Scaling MCP server as a Custom GPT action. Both paths give ChatGPT the ability to query your Scaling org over OAuth-secured MCP.

​
Server URL

https://mcp.scaling.cloud
1

Open Connectors

In ChatGPT, click your profile, then Settings → Connectors.
2

Add a custom connector

Click Add connector → Custom MCP server.
3

Paste the Scaling URL

Enter https://mcp.scaling.cloud and a friendly name like Scaling. Click Connect.
4

Authorize

ChatGPT redirects you to Clerk’s hosted OAuth consent screen. Sign in with your Scaling account and approve.
5

Done

The connector appears as Connected. You can now ask ChatGPT to use Scaling tools in conversations.

​
Path B: Custom GPT action (fallback)

If your ChatGPT plan does not yet expose Connectors, you can wire the MCP server in via a Custom GPT with an MCP-aware action. Detailed steps depend on the GPT Builder UI; the essential pieces are:
  • Authentication type: OAuth
  • Authorization URL: discovered from https://mcp.scaling.cloud/.well-known/oauth-authorization-server
  • Token URL: same source
  • Scopes: the default set returned by the server
  • Server URL: https://mcp.scaling.cloud
OpenAI’s Actions OAuth guide walks through the exact form fields.

​
Verify it’s working

Open Settings → Connections in the Scaling dashboard. You should see a row for ChatGPT (or your Custom GPT name) with a recent Last used timestamp.

​
Revoke access

Click Revoke in Settings → Connections to invalidate ChatGPT’s token immediately. Subsequent calls from that ChatGPT instance return 401 Unauthorized until you reconnect.

​
Limitations

  • Custom GPT actions cap each request response size, so very large incident lists may be truncated. Prefer Connector mode where available.
  • OpenAI rolls Connector availability out incrementally — not every plan or region has it yet. Use the Custom GPT path until your account exposes the Connector UI.