workingproxysites.

Provider integration · Official publisher

Bright Data MCP

Bright Data MCP combines source discovery, page extraction and optional browser tools. The main setup choice is where the server runs; the next is which tools the agent can see. Start with the capabilities your research task needs.

Server publisher: Bright Data. Guide maintained by Working Proxy Sites.

Documentation reviewed · Next review

What you can use it for

  • Collect search results before choosing pages to read.
  • Turn a public page into Markdown for a research workflow.
  • Use browser interactions when enabled by the selected tool mode.

Documented capabilities: Web search · Scraping & extraction · Browser automation.

Account & usage

Account required
Yes
Required product
Bright Data Web Unlocker API; additional products may be used by optional tools.
Billing basis
API usage draws from the account’s allowances; additional usage and optional tools follow the applicable product terms. Free server software does not make all API usage free. Provider pricing.
Server software license
MIT

When to choose Bright Data MCP

A useful fit when

  • Research that starts with a search query and continues into source pages.
  • Tasks that may need an interactive browser after basic page extraction.

Check another approach when

  • Anonymous access without a Bright Data account and API token.
  • A configuration that sets groups or tools without the browser group but expects browser tools.
Sources: Bright Data MCP overview, Bright Data tool reference, Bright Data remote quickstart

Connect to Bright Data MCP

Choose a connection profile below and replace credential placeholders in your own MCP client. The transport describes the MCP connection.

Hosted MCP

Streamable HTTP · hosted

Endpoint
https://mcp.brightdata.com/mcp
Authentication
Bright Data API token in the token query parameter.

Before you connect

  • Bright Data account and API token.
  • A client supporting remote Streamable HTTP MCP.

Cursor mcp.json (JSON) Sources: Bright Data remote quickstart, Bright Data official repository, Cursor MCP configuration reference

Cursor: .cursor/mcp.json (project) or ~/.cursor/mcp.json (global). Configuration guide.

{
  "mcpServers": {
    "bright-data": {
      "url": "https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN"
    }
  }
}

Local package 2.11.3 (default mode)

Local stdio · local

Package / runtime
@brightdata/[email protected]
Authentication
API_TOKEN environment variable.

Before you connect

  • Bright Data account and API token.
  • Node.js and npm available to the MCP client.

Cursor mcp.json (JSON) Sources: Bright Data official repository, Bright Data npm package 2.11.3, Cursor MCP configuration reference

Cursor: .cursor/mcp.json (project) or ~/.cursor/mcp.json (global). Configuration guide.

{
  "mcpServers": {
    "bright-data": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@brightdata/[email protected]"
      ],
      "env": {
        "API_TOKEN": "YOUR_API_TOKEN"
      }
    }
  }
}

Search, read and decide whether browser tools are needed

A suggested sequence using the documented tools. Check the connected server's tool schema before making a call.

  1. Find candidate sources

    Ask search_engine for sources on a concrete topic. Select the relevant result URLs before expanding the task.

    Tools: search_engine

    Sources: Bright Data tool reference

  2. Read the selected pages

    Pass those URLs to scrape_as_markdown and ask the agent to preserve source links with its findings. Review the returned material before drawing conclusions.

    Tools: scrape_as_markdown

    Sources: Bright Data tool reference

  3. Escalate to an interactive page when necessary

    If the task requires interacting with a page, use scraping_browser_navigate; browser tools are enabled by default unless groups or tools restrict the server. Navigation starts a browser workflow; it does not itself establish that the requested information was collected.

    Tools: scraping_browser_navigate

    Sources: Bright Data tool reference, Bright Data official repository

Selected tools & example tasks

  • search_engine

    Retrieve results from a supported search engine.

    Example task: Find relevant pages for a research question.

    Sources: Bright Data tool reference
  • scrape_as_markdown

    Read one page as Markdown.

    Example task: Extract a page’s text for a cited summary.

    Sources: Bright Data tool reference
  • scraping_browser_navigate

    Navigate a remote browser session (browser group; enabled by default unless groups/tools restrict the server).

    Example task: Open an interactive page as the first step in a browser workflow.

    Sources: Bright Data official repository, Bright Data tool reference

Troubleshoot the connection

A copied hosted configuration cannot authenticate.

Replace YOUR_API_TOKEN with a token from the Bright Data account used for the task. Keep the configured MCP URL and credential parameter intact.

Sources: Bright Data remote quickstart

Limits & undocumented behavior

  • All tools, including browser automation, are enabled by default. Set groups/tools (hosted URL) or GROUPS/TOOLS (local) to limit them. Pro mode was removed in 3.1.0, and PRO_MODE is now ignored.
  • The remote template contains a token placeholder in its URL; replace it only in your own client configuration.
  • The reviewed tool documentation does not establish proxy purchasing or proxy account management.

Bright Data MCP: questions & answers

Which Bright Data MCP connection should I start with?

Hosted is a practical starting point for a client with remote MCP support. Local stdio is useful when you need a local process or control over the package version. Both paths require account authentication.

Sources: Bright Data remote quickstart, Bright Data official repository
Does the basic setup include interactive browser automation?

Browser tools are enabled by default; they are absent only when groups/tools restrict the server without the browser group.

Sources: Bright Data tool reference
Does the pinned local package version also pin the hosted server?

No. The package version applies to the local command shown here. The hosted URL is a separate deployment, and this directory does not establish its deployed software revision.

Sources: Bright Data official repository, Bright Data npm package 2.11.3, Bright Data remote quickstart

Sources & review history

This listing is documentation-confirmed. We reviewed the primary sources below; this does not establish successful tool execution.

Record updated . Review due .

  1. Bright Data MCP overviewChecked · Publisher, scope and account-level usage allowance
  2. Bright Data remote quickstartChecked · Streamable HTTP endpoint, account and token authentication
  3. Bright Data tool referenceChecked · Search, page extraction, mode and group requirements
  4. Bright Data official repositoryChecked · Local package setup and browser navigation tool; CHANGELOG: all tools on by default from 3.1.0; PRO_MODE ignored
  5. Bright Data npm package 2.11.3Checked · Published package version and MIT license
  6. Cursor MCP configuration referenceChecked · Client JSON format and remote header support

Spotted a change? Send the source and the affected setup step to our contact address.

Alternatives to Bright Data MCP

These integrations share documented capabilities with Bright Data MCP. Compare their setup and product requirements for your task.

Evomi MCP

Inspect Evomi proxy products, targeting and usage, rotate sessions, and run scraping jobs through a local Python MCP server.

Shared capabilities
Web search · Scraping & extraction · Browser automation
Connection choices
Local stdio
Product to check
Evomi Public API access plus the proxy or Scraper API product used by the tool.
Compare setup & authentication →

Firecrawl MCP

Choose keyless, OAuth or API-key access to Firecrawl search and extraction tools. Check which operations need an account before connecting your client.

Shared capabilities
Web search · Scraping & extraction · Browser automation
Connection choices
Streamable HTTP
Product to check
Keyless hosted Search, Scrape and Parse; a Firecrawl account/team for authenticated tools.
Compare setup & authentication →

Novada MCP

Search, extract pages, use browser tools and configure proxy connections through Novada’s hosted endpoint or local npm package.

Shared capabilities
Web search · Scraping & extraction · Browser automation
Connection choices
Streamable HTTP · Local stdio
Product to check
Novada account with access to the scraping, browser or proxy product selected by the tool.
Compare setup & authentication →

Oxylabs MCP

Connect an agent to Oxylabs scraping and AI Studio tools, with separate credentials for each product.

Shared capabilities
Web search · Scraping & extraction · Browser automation
Connection choices
Streamable HTTP
Product to check
Oxylabs Web Scraper API for the included setup; AI Studio requires its own key.
Compare setup & authentication →