> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pipecorn.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Pipecorn MCP server

> Connect Claude, ChatGPT, Claude Code, Cursor, or any MCP-compatible assistant to Pipecorn and let it find companies, search leads, and enrich contacts from a chat.

The Pipecorn MCP server lets AI assistants and agents use Pipecorn directly
through the [Model Context Protocol](https://modelcontextprotocol.io). Once
connected, your assistant can find companies, search leads, enrich contacts,
manage Sales Navigator lists, and pull buying signals on your behalf, using
the same data and credits as the [REST API](/api-reference/main-use-cases).

|                    |                                                                                                                            |
| ------------------ | -------------------------------------------------------------------------------------------------------------------------- |
| **Server URL**     | `https://mcp.pipecorn.com/mcp`                                                                                             |
| **Transport**      | Streamable HTTP                                                                                                            |
| **Authentication** | OAuth (Claude, ChatGPT) or an `X-API-Key` header (Claude Code, Cursor, custom clients)                                     |
| **Tools**          | 31, across companies, leads, contacts, lists, personas, signals, jobs, and utilities. See the [tool reference](/mcp/tools) |
| **Discovery**      | [`/.well-known/mcp.json`](https://docs.pipecorn.com/.well-known/mcp.json)                                                  |

## Set up your client

<CardGroup cols={2}>
  <Card title="Claude" icon="message" href="/mcp/setup/claude">
    Add Pipecorn as a custom connector and sign in with OAuth.
  </Card>

  <Card title="ChatGPT" icon="comment" href="/mcp/setup/chatgpt">
    Create a Pipecorn app in developer mode with OAuth.
  </Card>

  <Card title="Claude Code" icon="terminal" href="/mcp/setup/claude-code">
    One command with your API key.
  </Card>

  <Card title="Cursor" icon="code" href="/mcp/setup/cursor">
    Add the server to your `mcp.json` with your API key.
  </Card>
</CardGroup>

Any other MCP client that supports custom headers can connect the same way
Claude Code and Cursor do: point it at the server URL and send your API key
in the `X-API-Key` header.

## Prerequisites

* A Pipecorn account with an [API key](/api-reference/API-key). OAuth clients
  sign in with your Pipecorn login instead and do not need the key.
* Node.js, only for clients that run the server through `mcp-remote`
  (Claude Code, Cursor).

## How it fits with the other surfaces

The MCP server exposes the same primitives as the REST API and the
[CLI](/cli/overview), so anything you can do in a chat you can also script.
Use MCP when a person is driving an assistant. Use the API or CLI when code
is driving the workflow. Credits and plan limits are shared across all three;
see [Credits vs limits](/api-reference/credits-vs-limits).

Before any large extraction or enrichment, the assistant should call the
`preview_*` tools and `get_credits` to size the job and confirm the balance.
The [agents guide](https://docs.pipecorn.com/agents.md) lists the conventions
we recommend agents follow.

## Roadmap

The MCP server is updated regularly. Upcoming work:

* Additional tools for Signals and Lists
* CSV download support
* Custom UI cards
* Token-usage optimizations