Data as of

Hyperbrowser MCP Server for Claude Desktop: Setup & Config

Claude Desktop: Hyperbrowser MCP Server combines web scraping, crawling, structured extraction, and cloud browser agents.

What you can do with Hyperbrowser in Claude Desktop

Pick it when an assistant needs both web extraction and a choice of cloud browser agents.

  • Extract page content as Markdown or screenshots
  • Crawl linked pages for content collection
  • Convert webpage content into schema-shaped data

Before you start: Hyperbrowser in Claude Desktop

  • npx
  • A Hyperbrowser API key
  • An MCP-compatible client

The README supplies client configurations for Cursor, Windsurf, and Claude Desktop. A source checkout can instead be built with npm and launched from dist/server.js.

Hyperbrowser Claude Desktop configuration

  1. Choose Claude Desktop configuration scope

    For Hyperbrowser MCP Server, open macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json. Merge under mcpServers via Settings → Developer → Edit Config when available.

  2. Add Hyperbrowser MCP Server to Claude Desktop

    Use the mcp-hyperbrowserai entry below in mcpServers. Keep Hyperbrowser MCP Server's documented stdio connection fields and your other Claude Desktop entries.

  3. Check Hyperbrowser MCP Server in Claude Desktop

    Hyperbrowser MCP Server: Fully quit and reopen Claude Desktop; check developer diagnostics.

Config file · macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

JSON
{
  "mcpServers": {
    "mcp-hyperbrowserai": {
      "command": "npx",
      "args": [
        "-y",
        "hyperbrowser-mcp"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<Hyperbrowser API key>"
      }
    }
  }
}

Hyperbrowser tools available to Claude Desktop

Hyperbrowser's stored tool list documents 10 tools; compare it with Claude Desktop's connected tool list, because the stored list is not a live check.

scrape_webpage

When Claude Desktop requests scrape_webpage from Hyperbrowser, review the action before approving it.

Retrieve one page as text, HTML, links, or screenshots with browser options.

crawl_webpages

When Claude Desktop requests crawl_webpages from Hyperbrowser, review the action before approving it.

Collect content from linked pages starting at a selected URL.

extract_structured_data

When Claude Desktop requests extract_structured_data from Hyperbrowser, review the action before approving it.

Extract webpage information according to a supplied schema.

browser_use_agent

When Claude Desktop requests browser_use_agent from Hyperbrowser, review the action before approving it.

Run explicit browser tasks with the Browser Use cloud agent.

openai_computer_use_agent

When Claude Desktop requests openai_computer_use_agent from Hyperbrowser, review the action before approving it.

Delegate general browser interactions to OpenAI computer use.

claude_computer_use_agent

When Claude Desktop requests claude_computer_use_agent from Hyperbrowser, review the action before approving it.

search_with_bing

When Claude Desktop requests search_with_bing from Hyperbrowser, review the action before approving it.

create_profile

When Claude Desktop requests create_profile from Hyperbrowser, review the action before approving it.

delete_profile

When Claude Desktop requests delete_profile from Hyperbrowser, review the action before approving it.

list_profiles

When Claude Desktop requests list_profiles from Hyperbrowser, review the action before approving it.

Hyperbrowser: full stored tool descriptions

Troubleshooting Hyperbrowser in Claude Desktop

  • Fully quit and reopen Claude Desktop for Hyperbrowser MCP Server. If npx hyperbrowser-mcp (Node.js/npm) is missing, check the desktop process's PATH, which may differ from your terminal.
  • In Claude Desktop's Hyperbrowser MCP Server entry, check HYPERBROWSER_API_KEY. Keep secrets out of shared screenshots and repositories.
  • Hyperbrowser MCP Server uses stdio here: Claude Desktop must keep the configured process running; protocol output belongs on stdout and diagnostics on stderr.
  • Known limitation: Requires a Hyperbrowser API key
  • Known limitation: The MCP toolset exposes only a subset of Hyperbrowser API features
  • General Claude Desktop troubleshooting lives in the /clients/claude-desktop guide.

Hyperbrowser MCP Server — details. Claude Desktop guide

Frequently asked questions

Claude Desktop loads Hyperbrowser MCP Server from claude_desktop_config.json: ~/Library/Application Support/Claude/ on macOS, or %APPDATA%\Claude\ on Windows. Open it via Settings → Developer → Edit Config when available.

Claude Desktop's mcp-hyperbrowserai example uses stdio.

Hyperbrowser MCP Server's Claude Desktop configuration invokes npx hyperbrowser-mcp (Node.js/npm).

Claude Desktop's Hyperbrowser MCP Server example references HYPERBROWSER_API_KEY. Check these names if startup or authentication fails; the example alone does not distinguish required from optional values.

Hyperbrowser MCP Server's stored list contains 10 tools, including scrape_webpage, crawl_webpages, extract_structured_data. Compare those names with Claude Desktop's connected tool list; the stored list is not a live check.

Browser Use, OpenAI computer use, and Claude computer use are exposed.

Yes. extract_structured_data uses a schema to shape the extracted webpage data.