Pieces MCP + ChatGPT Developer Mode | Setup

Integrate Pieces MCP with ChatGPT Developer Mode via ngrok or HTTPS proxy. Requires Pro, Plus, Business, Enterprise, or Education plan.

Pieces MCP integration with ChatGPT Developer Mode

Get Started

Integrating the Pieces MCP with ChatGPT Developer Mode brings your workflow context directly into ChatGPT in your browser. ChatGPT Developer Mode requires a remote HTTPS URL—localhost will not work. You must expose PiecesOS via ngrok or another HTTPS proxy first.

**Remote HTTPS required:** ChatGPT Developer Mode cannot use localhost. You must expose PiecesOS via ngrok or another HTTPS tunnel before adding the connector. **Plans required:** ChatGPT Developer Mode is available on Pro, Plus, Business, Enterprise, and Education plans. Status: Beta.

Prerequisites

There are three prerequisites for integrating Pieces with ChatGPT Developer Mode:

Ensure you have a Pro, Plus, Business, Enterprise, or Education ChatGPT plan. PiecesOS must be running locally. This is *required* for the MCP server to provide your workflow data. Expose PiecesOS via an ngrok tunnel to get a public HTTPS URL. The tunnel must be active when you add the connector and when you use ChatGPT. See [ngrok Setup](/products/mcp/ngrok-setup) for detailed instructions.

Installing PiecesOS & Configuring Permissions

Follow the instructions below for a detailed guide on setting up and configuring PiecesOS to correctly pass captured workflow context to the Pieces MCP server.

Setting Up ChatGPT Developer Mode

ChatGPT uses a Connectors UI to add MCP servers. You cannot add localhost URLs—only remote HTTPS endpoints.

Setup Steps

Open [chatgpt.com](https://chatgpt.com) in your browser. Click your **profile icon** → **Settings**. Go to **Connectors** in the sidebar. Scroll to the bottom and click `Advanced Settings`. Toggle `Developer Mode (beta)` ON. Return to **Connectors** and click `Create` (appears after enabling Developer Mode). * **Connector name:** `Pieces LTM` * **MCP Server URL:** `https://YOUR_NGROK_URL.ngrok-free.app/model_context_protocol/2025-03-26/mcp` * **Description:** (optional) `Search and retrieve from Pieces Long-Term Memory` Click `Create` to save the connector. The URL must point to the `/mcp` endpoint path. Subdirectory paths like `/functions/v1/mcp` do not work.

Using Pieces MCP Server in ChatGPT

Once integrated, you can utilize Pieces LTM directly in ChatGPT.

Start a new conversation in ChatGPT. Look for the **tools indicator** in the chat. Ask: *"What tools do you have from Pieces?"* Try: *"What did I work on yesterday?"* ChatGPT will use the `ask_pieces_ltm` tool to query your Long-Term Memory. Check out this [MCP-specific prompting guide](/products/mcp/prompting) if you want to effectively utilize the Long-Term Memory Engine (LTM-2.7) with your new Pieces MCP server.

Updating

To update the ngrok URL after a tunnel restart:

  1. Go to Settings > Connectors
  2. Find "Pieces LTM" and click Edit
  3. Update the MCP Server URL
  4. Save

Troubleshooting

If you're experiencing issues integrating Pieces MCP with ChatGPT Developer Mode:

  1. Cannot Use localhost: ChatGPT requires public HTTPS. Use ngrok or another HTTPS proxy.

  2. Connector Not Connecting: Ensure the ngrok tunnel is running and the URL is accessible in a browser.

  3. Authentication Required: Use no-auth mode; Pieces does not need OAuth.

  4. Tools Not Appearing: Refresh the page after adding the connector.

  5. Developer Mode Not Visible: Ensure you are on a supported plan (Pro/Plus/Business/Enterprise/Education).

  6. Write Actions Need Confirmation: By default, writes require manual approval in ChatGPT.

You're now set to enhance your ChatGPT workflow with powerful context retrieval through Pieces MCP. Happy coding!