Plotly Cloud MCP Server

You can connect Plotly Cloud to an AI assistant, such as Claude, Cursor, or ChatGPT. One connection lets your assistant manage your Dash apps on Plotly Cloud and search the official Dash documentation. Ask the assistant to build an app and publish it, find out why an app stopped working, or look up how a component works, and it does the work in Plotly Cloud for you.

The connection uses the Model Context Protocol (MCP), an open standard that lets AI assistants work with other services.

Note: If you previously connected your assistant to the Dash Docs MCP server, you no longer need that connection. The Plotly Cloud MCP server includes documentation search. Remove the old connection after you connect to Plotly Cloud.

Connecting Your Assistant

The MCP server address is:

<a href="https://cloud.plotly.com/_mcp">https://cloud.plotly.com/_mcp</a>

Every connection requires a one-time sign-in with your Plotly Cloud account in the browser. During sign-in, you also choose the team that the connection works with. See Teams for details.

Don’t have a Plotly Cloud account yet? Create one for free.

Select your assistant for setup instructions.

Claude

Plotly Cloud is available in the Claude connector directory, so you don’t need to enter the server address.

  1. In Claude (web or desktop), open the connector directory and find Plotly Cloud.
  2. Select Connect.
  3. Sign in with your Plotly Cloud account.
  4. Choose the team you want to work with.

Claude Code

Run the following command in your terminal to add the server for all your projects:

claude mcp add plotly-cloud --transport http --scope user <a href="https://cloud.plotly.com/_mcp">https://cloud.plotly.com/_mcp</a>

Restart Claude Code so that it loads the new server. Then type /mcp in Claude Code, select the server, and select Authenticate to sign in with your Plotly Cloud account in the browser.

Cursor

Open your Cursor MCP settings and add the following:

{
  "mcpServers": {
    "plotly-cloud": {
      "type": "http",
      "url": "https://cloud.plotly.com/_mcp"
    }
  }
}

When you first use a tool, Cursor opens a browser window so you can sign in with your Plotly Cloud account.

Windsurf

Open your Windsurf MCP config file and add the following:

{
  "mcpServers": {
    "plotly-cloud": {
      "type": "http",
      "url": "https://cloud.plotly.com/_mcp"
    }
  }
}

When you first use a tool, Windsurf opens a browser window so you can sign in with your Plotly Cloud account.

ChatGPT

ChatGPT supports custom remote MCP servers in developer mode (available on eligible plans).

  1. In ChatGPT, go to Settings > Connectors and turn on developer mode.
  2. Add a new connector with the name Plotly Cloud and the URL https://cloud.plotly.com/_mcp.
  3. Sign in with your Plotly Cloud account when prompted.

Gemini CLI

Add the following to your ~/.gemini/settings.json:

{
  "mcpServers": {
    "plotly-cloud": {
      "httpUrl": "https://cloud.plotly.com/_mcp"
    }
  }
}

Then start Gemini CLI and authenticate:

/mcp auth plotly-cloud

A browser window opens so you can sign in with your Plotly Cloud account. For general setup, see the Gemini CLI MCP documentation.

Codex CLI

Add the server from your terminal:

codex mcp add plotly-cloud --url <a href="https://cloud.plotly.com/_mcp">https://cloud.plotly.com/_mcp</a>

Sign in with your Plotly Cloud account when prompted. For general setup, see the Codex CLI MCP documentation.

Kiro

Open the Command Palette (Cmd+Shift+P on Mac, Ctrl+Shift+P on Windows/Linux) and search for “MCP”, or go to File > Settings to open your mcp.json. Then add:

{
  "mcpServers": {
    "plotly-cloud": {
      "type": "http",
      "url": "https://cloud.plotly.com/_mcp"
    }
  }
}

Save the file. When the server connects, sign in with your Plotly Cloud account in your browser. For general setup, see the Kiro MCP documentation.

Replit

In your Replit workspace, go to Integrations > MCP Servers and add a new server with the name Plotly Cloud and the URL https://cloud.plotly.com/_mcp.

Sign in with your Plotly Cloud account when prompted. For general setup, see the Replit MCP documentation.

Lovable

In Lovable, open Settings > Connectors and add a custom MCP server with the URL https://cloud.plotly.com/_mcp.

Sign in with your Plotly Cloud account when prompted. For general setup, see the Lovable MCP documentation.

Figma Make

In a Figma Make file, select the + icon, then Connectors, and add a custom MCP server with the URL https://cloud.plotly.com/_mcp.

Sign in with your Plotly Cloud account when prompted. For general setup, see Connect external tools to Figma Make using MCP.

Other MCP clients

Add a remote (HTTP) MCP server to your client with the URL https://cloud.plotly.com/_mcp. Your client might call this a server, a connector, or an integration. Check your client’s documentation for where to add it.

Sign in with your Plotly Cloud account when prompted.

To confirm that the connection works, ask your assistant: “Which Plotly Cloud team am I connected to?”

What Your Assistant Can Do

After you connect, you only need to describe what you want and it works out the steps to do it.

Your assistant can build a Dash app from a description or from data you share, and publish it to Plotly Cloud. For an existing app, it can read the app’s current code, make the changes you ask for, and publish a new version. It can also list your apps and check whether they’re running.

When an app fails to publish or stops working, your assistant can read the app’s logs, explain what went wrong, and fix it. You can also ask it to start, stop, restart, or delete apps.

Your assistant can search and read the Dash documentation, whether or not you’re working on an app. Instead of relying on what it remembers about Dash, it reads the current documentation pages as it answers your questions and writes code.

Example Prompts

Teams

When you connect, you can only choose one team to work within. Meaning your assistant publishes new apps to it and can only list, update, and manage the apps in it.

To work with a different team, disconnect the Plotly Cloud MCP server in your assistant, and then connect again and choose the new team when you sign in. You don’t need to sign out of Plotly Cloud.

Security and Permissions

Your assistant acts as you. It can only do what your Plotly Cloud account is allowed to do. For example, if you can’t edit an app in Plotly Cloud, your assistant can’t either. Connections always require signing in with a Plotly Cloud account, and you can’t use API keys to connect.

New apps follow your team’s settings. For example, whether a new app is public or private depends on your team’s default. Your plan’s limits also apply in the same way as in Plotly Cloud, such as how many apps can run at the same time or whether you can choose a custom app URL.

Deleting an app is permanent, and you can’t recover a deleted app. Assistants are instructed to check with you before they delete an app, but review these requests carefully.

Enterprise Library Documentation

Documentation for open-source Dash is available to all Plotly Cloud accounts. Documentation for enterprise libraries, such as Dash Design Kit, Snapshot Engine, and Dash Embedded, requires a Plotly Cloud subscription with an Enterprise plan.

Note that the Plotly Cloud MCP server doesn’t yet support Dash Enterprise sign-in. You will need an Enterprise Cloud account, which can be provided by your Plotly account manager.

Plotly Cloud MCP vs. Dash App MCP Servers

The Plotly Cloud MCP server is different from the MCP server that an individual Dash app can provide. The Plotly Cloud MCP server manages your apps and searches the documentation. It doesn’t give your assistant access to the data or callbacks inside your apps.

To let an assistant work with the data in one app, turn on that app’s MCP server and connect to it separately. See Dash MCP on Plotly Cloud. Your assistant can give you an app’s MCP server address if you ask for it.

If Your Team Reaches Its App Limit

If your team is already running as many apps as its plan allows, your assistant will tell you during the publishing process. To publish the app, stop or delete another app, or upgrade your plan. Then ask your assistant to try again.

Troubleshooting

Problem Solution
Apps are published to the wrong team Disconnect Plotly Cloud in your assistant, connect again, and choose the correct team.
Your assistant says your team can’t be found The team might have been deleted, or you might no longer be a member. Disconnect, connect again, and choose a team.
Your assistant reports a temporary problem Wait a moment and try again.
An app doesn’t go live because of a limit See If Your Team Reaches Its App Limit.
Your assistant can’t use the app URL you asked for Custom app URLs depend on your plan. Let Plotly Cloud choose the URL instead.
Your assistant can’t find or change an app Check that the app is in the team you chose when you connected, and that you have permission to edit it in Plotly Cloud.

Resources