Hyperbrowser MCP Server

Name: Hyperbrowser MCP Server
Author: Hyperbrowser

Scrape, crawl, search and run browser agents through Hyperbrowser's cloud browsers

Official · Hyperbrowser TypeScript Local · stdio Browser automation Web search

GitHub npm Website

34 popularity score ★ 793 ↓ 508/wk

Connect your client

Local (npm)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.

Or add it by command or config

It needs HYPERBROWSER_API_KEY. Replace each <NAME> with your value.

Claude CodeTerminal

claude mcp add hyperbrowser -e HYPERBROWSER_API_KEY=<HYPERBROWSER_API_KEY> -- npx -y hyperbrowser-mcp

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "hyperbrowser": {
      "command": "npx",
      "args": [
        "-y",
        "hyperbrowser-mcp"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<HYPERBROWSER_API_KEY>"
      }
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "hyperbrowser": {
      "command": "npx",
      "args": [
        "-y",
        "hyperbrowser-mcp"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<HYPERBROWSER_API_KEY>"
      }
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "hyperbrowser": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "hyperbrowser-mcp"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<HYPERBROWSER_API_KEY>"
      }
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.hyperbrowser]
command = "npx"
args = ["-y", "hyperbrowser-mcp"]

[mcp_servers.hyperbrowser.env]
HYPERBROWSER_API_KEY = "<HYPERBROWSER_API_KEY>"

Gemini CLITerminal

gemini mcp add --scope user -e HYPERBROWSER_API_KEY=<HYPERBROWSER_API_KEY> hyperbrowser npx -- -y hyperbrowser-mcp

Local (from source)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio

Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.

Or add it by command or config

It needs HYPERBROWSER_API_KEY. Replace each <NAME> with your value.

FirstTerminal

git clone https://github.com/hyperbrowserai/mcp.git hyperbrowser-mcp
cd hyperbrowser-mcp
npm install
npm run build

Claude CodeTerminal

claude mcp add hyperbrowser -e HYPERBROWSER_API_KEY=<HYPERBROWSER_API_KEY> -- node '<ABSOLUTE_PATH_TO_CLONE>/dist/server.js'

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "hyperbrowser": {
      "command": "node",
      "args": [
        "<ABSOLUTE_PATH_TO_CLONE>/dist/server.js"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<HYPERBROWSER_API_KEY>"
      }
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "hyperbrowser": {
      "command": "node",
      "args": [
        "<ABSOLUTE_PATH_TO_CLONE>/dist/server.js"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<HYPERBROWSER_API_KEY>"
      }
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "hyperbrowser": {
      "type": "stdio",
      "command": "node",
      "args": [
        "<ABSOLUTE_PATH_TO_CLONE>/dist/server.js"
      ],
      "env": {
        "HYPERBROWSER_API_KEY": "<HYPERBROWSER_API_KEY>"
      }
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.hyperbrowser]
command = "node"
args = ["<ABSOLUTE_PATH_TO_CLONE>/dist/server.js"]

[mcp_servers.hyperbrowser.env]
HYPERBROWSER_API_KEY = "<HYPERBROWSER_API_KEY>"

Gemini CLITerminal

gemini mcp add --scope user -e HYPERBROWSER_API_KEY=<HYPERBROWSER_API_KEY> hyperbrowser node -- '<ABSOLUTE_PATH_TO_CLONE>/dist/server.js'

Settings 2

`HYPERBROWSER_API_KEY`	Required: Hyperbrowser API key; every tool call needs it. HB_API_KEY is read first if set.
`SSE_PORT`	Optional: Port for the SSE mode started with the --sse flag; 3001 by default.

What it does

Hyperbrowser MCP Server is the Hyperbrowser company's own bridge between an MCP client and its hosted browser service. The server runs on your machine as a small Node program, but every page it opens is loaded in a browser session on Hyperbrowser's cloud, billed to your Hyperbrowser API key.

It covers three kinds of work. Fetching: scrape one page into markdown or a screenshot, crawl a site page by page, or have a page turned into JSON that follows a schema you give. Searching: run a Bing web query through the service. Delegating: hand a plain-language task to one of three browser agents (Browser Use, OpenAI's computer-use model, or Claude computer use) that click and type on their own in a cloud browser. Three more tools create, list and delete persistent browser profiles, so logged-in state can be kept between sessions.

In the published 1.0.25 code the caps sit in the tool schemas rather than the README: a crawl visits 10 pages by default and at most 100 (maxPages); each agent takes 25 steps by default and at most 100 (maxSteps); a Bing search returns 10 results by default, up to 50. When a session uses a profile, changes are saved back to it unless persistChanges is set to false. The server also serves Hyperbrowser's documentation as MCP resources.

What it needs

Runtime	Node.js 18+

Node.js 18 or newer (npx). A Hyperbrowser account and API key, set as HYPERBROWSER_API_KEY (HB_API_KEY is also read). Scraping, crawling, search and agent runs use Hyperbrowser's paid cloud service and are counted against that account.

Limits

Every tool calls Hyperbrowser's cloud; nothing is rendered locally, and without an API key each tool call fails. The README's manual command passes the key as an argument (npx hyperbrowser-mcp <KEY>), but the 1.0.25 code only reads it from the environment (src/utils.ts), so set the variable instead. Crawls stop at 100 pages and agent runs at 100 steps. The README notes the Hyperbrowser API offers more features than the server exposes. The repository's main branch is at 1.1.0, which is not published on npm; this page describes 1.0.25.

Who it suits

Suits developers who already pay for Hyperbrowser, or want cloud browsers with proxies, stealth mode and captcha solving, and need an assistant to scrape, crawl or hand off whole browsing tasks without running a browser on their own machine. Its agent tools suit multi-step jobs where the assistant should not drive each click. Those who want a free, local browser they control directly, or who cannot send page traffic through a third-party service, should pick a locally run browser server instead.

Tools 10

scrape_webpage: Fetch one URL and return its content as markdown, HTML, links or a screenshot
crawl_webpages: Follow links from a start URL and return each page's content, 10 pages by default
extract_structured_data: Turn one or more pages into JSON shaped by a prompt and an optional schema
search_with_bing: Run a Bing web search and return up to 50 results
browser_use_agent: Give a task to the Browser Use agent in a cloud browser
openai_computer_use_agent: Give a task to OpenAI's computer-use agent in a cloud browser
claude_computer_use_agent: Give a task to the Claude computer-use agent in a cloud browser
create_profile: Create a persistent browser profile on Hyperbrowser
delete_profile: Delete a persistent browser profile by its ID
list_profiles: List the account's persistent browser profiles, with paging

Popularity

34 popularity score

★ 793 ↓ 508/wk

GitHub stars	793	34.8 points
Stars in 30 days	+4	3.5 points
npm downloads, last week	508	21.7 points
Official server	Yes	8 points
Activity	No change in a year	× 0.5

Measured 2026-11-07. How the score works.

Where it is published

Code	hyperbrowserai/mcp ★ 793
npm	hyperbrowser-mcp 1.0.25, 2025-04-13

What it can reach

Runs	On your machine
Reaches	Drives a browser, The internet, Your account at a service
Changes things	Yes: it can write, send, create or delete
Sign-in	An API key or token
Licence	ISC (npm)
Last code change	2025-05-23

The server runs locally, but the URLs, tasks and prompts you give it are sent to Hyperbrowser's API, where cloud browsers load the pages; the agent tools let a third-party model (Browser Use, OpenAI or Anthropic's computer use) click and type on sites on its own for up to 100 steps. Agents and saved profiles can carry logged-in sessions, and create_profile and delete_profile change data in your Hyperbrowser account. Sign-in is one Hyperbrowser API key read from the environment; usage is billed to that account. Optional session settings turn on proxies, stealth mode and captcha solving. The --sse mode listens on port 3001 by default.

Over time

GitHub stars 793 +4 (+1 %)

793789

9 Oct 20267 Nov 2026

npm downloads a week 508 −1.4k (−74 %)

5.5k270

20 Oct 20257 Nov 2026

Before the site measured it, read from npm's daily download counts.

Popularity score 34 −25 (−42 %)

5931

9 Oct 20267 Nov 2026

What is rising across the directory

Sources

Page checked 2026-10-14.

Badge for your README

Shows this server's rank in its category, and stars for a server that runs on your machine. It updates every hour.

MarkdownREADME.md

[![Hyperbrowser MCP Server on EveryMCPServer](https://everymcpserver.com/badge/mcp-182.svg)](https://everymcpserver.com/servers/mcp-182)

Connect your client

Add in one click

Or add it by command or config

Add in one click

Or add it by command or config

Settings 2

What it does

What it needs

Limits

Who it suits

Tools 10

Similar servers

SearXNG MCP Server

Firecrawl MCP Server

Browser Use Cloud MCP

Browser Use

Exa MCP Server

Tavily MCP Server