> For the complete documentation index, see [llms.txt](/llms.txt)

# MCP Overview

# MCP Overview

The [Model Context Protocol](https://modelcontextprotocol.io/) (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Browserless provides two MCP servers, each designed for a different use case.

## Browserless MCP Server

**The primary MCP server for browser automation.** Connect Claude Desktop, Cursor, VS Code, Windsurf, or any MCP-compatible client and give your AI assistant the ability to scrape pages, take screenshots, generate PDFs, download files, and run custom Puppeteer code. All powered by Browserless's managed, stealth-enabled browser infrastructure.

**Use this when you need your AI to interact with the live web**: scraping content, extracting data, exporting pages, or executing browser-side JavaScript.

Key capabilities:

- **Browser Agent** — a stateful, snapshot-driven browser your AI can navigate and interact with across turns (logins, multi-step flows, form filling)
- **Smart Scraper** — cascading scrape strategies (HTTP fetch, proxy, headless browser, CAPTCHA solving) selected automatically
- **Custom code execution** — run arbitrary Puppeteer JavaScript on the Browserless cloud
- **File downloads** — trigger and retrieve browser-initiated downloads
- **Page export** — export any URL as HTML, PDF, or a full offline ZIP archive
- **Lighthouse performance audits** — run accessibility, SEO, performance, PWA, and best-practices audits on any URL
- **Website crawling** — crawl an entire site from a seed URL with configurable depth, path filters, and scrape options

[Get started with the Browserless MCP Server →](/mcp/browserless-mcp-server/setup)

## Browserless Docs MCP

A lightweight, read-only MCP server that gives your AI assistant direct access to the Browserless documentation. No API token required.

**Use this when you need your AI to answer questions about Browserless**: how to configure connections, which API endpoints to call, how to set up proxies, or any other documentation lookup.

Key capabilities:

- **Context-aware answers** — ask questions about Browserless APIs, BrowserQL, configuration, and more
- **Zero configuration** — no API token or authentication needed
- **Works everywhere** — supports Cursor, VS Code, Claude Desktop, ChatGPT Desktop, Claude Code, and any MCP-compatible client

[Get started with the Browserless Docs MCP →](/mcp/browserless-docs-mcp)

Most users will want the **Browserless MCP Server** for its full browser automation capabilities. Add the **Browserless Docs MCP** alongside it if you also want your AI assistant to have instant access to Browserless documentation while coding.

## Using Claude Code?

If you specifically use [Claude Code](https://docs.anthropic.com/en/docs/claude-code), there is also a dedicated [**Browserless Claude Code Plugin**](/ai-integrations/claude-code-plugin). It provides a slash-command interface (`/browserless:smart-scrape`, `/browserless:screenshot`, etc.) backed by the Browserless REST APIs rather than MCP. Install it from inside Claude Code:

```bash
/plugin marketplace add browserless/claude-plugin
/plugin install browserless@browserless
```

## FAQ & Troubleshooting

<details>
<summary>What is MCP in the context of Browserless?</summary>

MCP (Model Context Protocol) lets AI agents interact with Browserless through a standardized interface. The Browserless MCP server exposes browser automation tools that AI models can call directly.

</details>

<details>
<summary>Which AI agents work with the Browserless MCP server?</summary>

The Browserless MCP server works with any MCP-compatible client, including Claude Desktop, Claude Code, Cursor, Windsurf, and other AI tools that support the Model Context Protocol standard.

</details>

<details>
<summary>Do I need a Browserless account to use the MCP server?</summary>

The Browserless MCP Server requires an API token for browser automation tasks. Sign up for a free account at browserless.io to get your token. The Docs MCP server, which provides documentation access only, does not require a Browserless token.

</details>

## Next steps

  <a href="/ai-integrations/start" className="next-step-card-link">
    
      <h3 className="next-step-card-title">AI integrations overview</h3>
      all supported AI frameworks and tools
    
  </a>
  <a href="/browserql/start" className="next-step-card-link">
    
      <h3 className="next-step-card-title">BQL overview</h3>
      use BrowserQL for advanced browser automation
    
  </a>