Home / Help Articles Library / Connecting the MetaField MCP to ChatGPT

Help Article

Connecting the MetaField MCP to ChatGPT

This guide walks you through connecting the MetaField MCP to ChatGPT so you can query your MetaField data and workflows directly from your ChatGPT conversations.

Before you start

Make sure you have:

  • A ChatGPT account
  • Your MetaField MCP server URL:
    • US customers: https://us.mcp.metafield.com/mcp
    • Canadian customers: https://ca.mcp.metafield.com/mcp
  • Admin or Owner access, if you're setting this up for a Business or Enterprise organization

Note for Business and Enterprise admins: Setting this up for your whole organization works a little differently. See this OpenAI support article for how admins connect and publish a custom MCP server for the workspace.

Enable Developer mode

The ChatGPT Security and login settings with the Developer mode option
  1. In your browser, go to chatgpt.com.
  2. Click your profile icon in the bottom left.
  3. Select Settings > Security and login from the menu.
  4. Enable Developer mode.

Add the MetaField plugin

The ChatGPT Plugins page with the plus icon in the upper right
  1. Go to Settings > Plugins > Browse plugins.
  2. In the upper right, click the + icon to add a new plugin.
The new plugin window with name and MCP server URL fields
  1. Enter a name for the plugin, such as MetaField MCP.
  2. Enter your MetaField MCP server URL:
    • US customers: https://us.mcp.metafield.com/mcp
    • Canadian customers: https://ca.mcp.metafield.com/mcp
  3. Click Create, then Sign in.

Sign in with MetaField

The MetaField sign-in screen with username, password, and SSO options
  1. At the sign-in screen, log in with your MetaField username and password, or click Sign in with SSO.
  2. With either option, enter your customer alias.
  3. Return to ChatGPT.

Tip: Your customer alias is the text in your MetaField URL just before .agileframe.net. For example, if your URL is qatest.agileframe.net, enter qatest.

Optional: Under Settings > Plugins > Permissions, set the default to Allow low-risk actions. This reduces the number of permission checks ChatGPT asks for when you use the MCP.

Start using the MCP

A ChatGPT conversation using the MetaField plugin

The MCP should now be enabled for your account. Here are a few helpful use cases and questions you can ask when chatting in ChatGPT:

Use caseTry asking
Check soil compaction pass rates"What's the soil compaction pass rate on this project?"
Spot unassigned work"Which work orders next week don't have a technician assigned yet?"
Stay ahead of concrete breaks"What concrete breaks are due this week, and are any past their 28-day window?"
Get a full lab testing summary"Give me the lab testing rollup for this project: what's complete, what's outstanding, and what failed spec?"
See daily project activity"What happened on my projects yesterday, and what's scheduled for tomorrow?"
Track a sample through its lifecycle"Where is this sample in testing, and what's the status of every specimen and test on it?"
Compare technician time"How do hours logged compare to hours scheduled for each technician this week?"
Catch concrete that missed strength"Which cylinders came in below their design strength on this project?"
See a technician's schedule"What's scheduled for this technician tomorrow?"
Roll up boring testing status"What's the testing status for Boring B-12, all the way down through its samples, specimens, and tests?"

Go further with plugins and skills (optional)

A ChatGPT plugin can bundle skills with connected apps like MetaField. Skills are reusable instructions, such as how your firm formats a lab testing summary, and ChatGPT uses them automatically when they're relevant.

OpenAI has articles that walk you through how it all works:

Note: What's available depends on your ChatGPT plan and workspace settings.

Need help?

If you run into issues, contact your MetaField administrator or reach out to MetaField support.

Want to keep learning?

Browse more how-to articles and MetaField guides. You can also access in-app guidance using the Help Desk.

‹ MetaField MCPConnecting the MetaField MCP to Microsoft Copilot ›