---
agentTools:
  projectIndex: https://hub.phantombuster.com/llms.txt
---

# MCP server

Give AI tools a secured access to your PhantomBuster automations and data

The PhantomBuster MCP (Model Context Protocol) server gives AI assistants like ChatGPT, Claude, and Cursor a secure, standardized way to work with your PhantomBuster data and automations directly from a conversation.\
Instead of switching between PhantomBuster and your AI tool, you can ask your assistant to launch a Phantom, pull results, or build a filtered lead list, and it talks to PhantomBuster on your behalf.

## What is the PhantomBuster MCP server?

MCP is an open standard that lets AI agents connect to external services through a single, consistent interface. Our MCP server is centrally hosted and managed by PhantomBuster, exposes a curated subset of the PhantomBuster API as tools, and is protected by OAuth. It works with any MCP-compatible client and supports the Streamable HTTP transport.

When you connect, you'll **log in to your PhantomBuster account** and **choose the workspace you want to use**. The MCP server only ever acts within that selected workspace, to work in a different workspace you'll need to reconnect.

## What can you do with it?

The MCP server lets your AI assistant interact with the core of your selected PhantomBuster workspace. At a high level, it can:

* **Run and manage your automations** — launch Phantoms, check what's currently running, stop them, and adjust their configuration.
* **Retrieve results and outputs** — pull the data and logs produced by your Phantoms so your assistant can summarize, analyze, or act on them.
* **Work with your leads** — search across your stored leads using rich filters, and save or update records.
* **Build and maintain dynamic lists** — create or update filtered lead lists (the same dynamic lists you'd build in the app) straight from a prompt.
* **Explore your workspace** — look up your agents, scripts, and organization resources and usage to get context before taking action.

This is a curated set of capabilities focused on the most common workflows, not the entire PhantomBuster API. The available tools evolve over time, and your assistant will discover the current list automatically once connected.

## Connecting to the server

The MCP server is available at:

```
https://mcp.phantombuster.com
```

Authentication uses **OAuth**: the first time your client connects, you'll be redirected to sign in to PhantomBuster and authorize access. No API keys to copy or paste. After that, your assistant will reuse this connection.

### ChatGPT

[Add a custom connector](https://chatgpt.com/#settings/Connectors) (in **Settings → Apps**) pointing to the MCP URL:

```
https://mcp.phantombuster.com
```

Reference: <https://developers.openai.com/apps-sdk/deploy/connect-chatgpt>

### Claude

[Add a custom connector](https://claude.ai/customize/connectors) (in **Customize → Connectors**), pointing to the MCP URL:

```
https://mcp.phantombuster.com
```

Reference: <https://claude.com/docs/connectors/custom/remote-mcp>

### Claude Code CLI

Run the following command:

```bash
claude mcp add --transport http phantombuster https://mcp.phantombuster.com
```

Then, open a Claude session and type:

```
/mcp
```

Find the`phantombuster` MCP server, press `Enter` and then choose `Authenticate`

Reference: <https://code.claude.com/docs/en/mcp>

### Gemini CLI

Run the following command:

```bash
gemini mcp add --scope user --transport http phantombuster https://mcp.phantombuster.com
```

Then, open a Gemini session and type:

```
/mcp auth phantombuster
```

Reference: <https://geminicli.com/docs/tools/mcp-server/#how-to-set-up-your-mcp-server>

### Codex CLI

Run the following command:

```bash
codex mcp add phantombuster --url https://mcp.phantombuster.com
```

Reference: <https://developers.openai.com/codex/mcp>