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

# Connect peeps to any MCP-compatible AI tool or agent

> Use the peeps MCP server with any MCP-compatible AI tool or agent framework. Supports stdio and HTTP/SSE transports for local and cloud use.

peeps supports any MCP-compatible client through two transport options: **stdio** for local desktop apps and developer tools, and **HTTP/SSE** for cloud-hosted agents, server-side applications, and any client that connects to remote MCP servers. If your AI tool or framework speaks MCP, you can connect it to peeps.

## Stdio transport

Use stdio when you're running a local application, CLI tool, or development environment that spawns the MCP server as a child process. The client manages the server lifecycle — it starts `npx` for you and communicates over standard input/output.

**When to use:** Claude Desktop, Cursor, local agent scripts, any MCP client that accepts a `command` + `args` config.

Set the `PEEPS_API_KEY` environment variable to authenticate. Here's a generic stdio config:

```json theme={null}
{
  "mcpServers": {
    "peeps": {
      "command": "npx",
      "args": ["-y", "@peeps-ai/mcp"],
      "env": {
        "PEEPS_API_KEY": "your-api-key-here"
      }
    }
  }
}
```

Replace `your-api-key-here` with your key from [app.peeps.ai](https://app.peeps.ai) **Settings → API Keys**. The exact location of this config block depends on your client — consult its documentation for where to place MCP server definitions.

## HTTP/SSE transport

Use HTTP/SSE when you're running a cloud-hosted agent, a server-side application, or any client that connects to remote (rather than locally spawned) MCP servers. No local Node.js installation is required.

**When to use:** cloud agents, hosted AI platforms, Claude.ai web, LangChain/agent frameworks, any client that accepts a remote server URL.

| Setting        | Value                                         |
| -------------- | --------------------------------------------- |
| Server URL     | `https://mcp.peeps.ai/sse`                    |
| Authentication | `Authorization: Bearer <your-api-key>` header |

A generic HTTP/SSE config looks like this:

```json theme={null}
{
  "mcpServers": {
    "peeps": {
      "url": "https://mcp.peeps.ai/sse",
      "headers": {
        "Authorization": "Bearer your-api-key-here"
      }
    }
  }
}
```

The exact field names (`url`, `transport`, `headers`) vary by client — check your tool's MCP documentation for its specific schema.

## LangChain and agent frameworks

If you're building an agent with LangChain or a similar framework, you can connect to the peeps remote MCP server using the `langchain-mcp-adapters` package. Install it with `pip install langchain-mcp-adapters`, then connect like this:

```python theme={null}
from langchain_mcp_adapters.client import MultiServerMCPClient

client = MultiServerMCPClient({
    "peeps": {
        "url": "https://mcp.peeps.ai/sse",
        "headers": {
            "Authorization": "Bearer your-api-key-here"
        }
    }
})

tools = await client.get_tools()
```

Once you have `tools`, pass them to your LangChain agent or chain as you would any other tool list. The peeps tools (`search_people`, `get_person`, `search_organizations`, `get_organization`, `list_contacts`) will be available for the agent to call automatically.

<Note>
  The [Model Context Protocol specification](https://modelcontextprotocol.io) lists all officially supported client implementations and transport options. If your tool is MCP-compatible and isn't covered here, the stdio or HTTP/SSE configs above should work with minimal adaptation.
</Note>

## Guides for specific clients

<CardGroup cols={2}>
  <Card title="Claude" icon="comment" href="/integrations/claude">
    Step-by-step setup for Claude Desktop and Claude.ai, including config file locations and example prompts.
  </Card>

  <Card title="Cursor" icon="code" href="/integrations/cursor">
    Add peeps to Cursor for AI-powered people search right inside your code editor.
  </Card>
</CardGroup>