OfficialVerified

Zenrows MCP Server

Web data infrastructure access for AI assistants, agents, and applications.

Data last scanned today · Reviewed today

668 tools★ 19TypeScriptMIT

Overview

The Zenrows MCP server bridges AI models with web infrastructure, enabling real-time data retrieval from protected websites. By offloading headless browser orchestration, proxy rotation, and anti-bot management to Zenrows, it allows AI to interact with live web content through standard natural language prompts rather than custom scraping scripts. Designed for integration into any Model Context Protocol-compliant client, the server offers both hosted and locally configured deployment options. It uses standard authentication methods, such as API keys passed via Bearer tokens, to facilitate secure connections to web content infrastructure.

Our verdict

This server carries a high risk level with a security score of 38/100. The assessment is limited because critical aspects such as the authentication method and read-only mode support have not yet been editorially reviewed. Given the lack of formal audit, it is recommended primarily for users comfortable managing third-party infrastructure integrations where security-sensitive access patterns are not yet fully documented.

Supports proxy rotation and headless browser orchestration
Maintains a small supply-chain with only four direct dependencies
Uses MIT licensing for open access

Community-maintained rather than official vendor software
Security architecture has not undergone an editorial review
No verification for read-only mode behavior

Setup

Claude Desktop

{
  "mcpServers": {
    "zenrows-mcp": {
      "args": [
        "-y",
        "@zenrows/mcp"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
scrape	Fetches full-page content as Markdown, plain text, HTML, PDF, or screenshot.	low
extract	Extracts structured JSON content from web pages using automatic parsing or CSS selectors.	low
batch_create	Initiates a cloud batch request.	medium
batch_status	Checks the status of a cloud batch request.	low
batch_results	Retrieves the results of a completed cloud batch request.	low
batch_cancel	Cancels an active cloud batch request.	medium
batch_wait	Waits for a cloud batch request to complete.	low
browser_*	Provides over 30 tools for browser automation including navigation, clicks, form interaction, JS execution, cookies, and session management.	medium

Compatibility

Client	Local	Docker	Remote	Read-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

›What authentication methods does the remote Zenrows MCP server support?

The remote server supports authentication via OAuth or an API key provided as a Bearer token in the Authorization header.

›Does the remote MCP server provide an account for me automatically?

No, the remote server does not auto-create accounts; you must use OAuth to create a free account or provide an existing API key.

›How does the local stdio version handle authentication if I do not provide an API key?

If no key is provided, the stdio server automatically performs an auto-signup to provision a free plan account, persists the credentials to your local file system, and outputs a claim URL.

›Can I disable the auto-signup feature in the local MCP server?

Yes, you can opt out of the auto-signup behavior by setting the environment variable ZENROWS_AUTO_SIGNUP to false.

›What are the system requirements for running the local version of this MCP server?

The local server requires Node.js to be installed on your machine so that npx can execute the package.