Bright Data's official MCP server gives an AI client robust web access — search engines, page extraction and unblocking for sites that resist plain fetches. It's a go-to when a basic fetch tool keeps hitting walls. Here's the setup.
Step 1 — get an API token
Create a Bright Data account and generate an API token in the dashboard. The MCP has a free tier for basic tools; advanced unblocking/browser tools consume credits.
Step 2 — configure the client
{
"mcpServers": {
"brightdata": {
"command": "npx",
"args": ["-y", "@brightdata/mcp"],
"env": {
"API_TOKEN": "your_bright_data_token"
}
}
}
}
Restart the client; see MCP config file location. Bright Data also offers a hosted remote endpoint for clients that support remote MCP.
What it gives your AI
Search the web, fetch and clean pages as Markdown, and use unblocking/browser tools for sites that block scrapers — useful for research agents that need current, reliable data.
Verify
Ask: "Search for the latest MCP directory comparisons and extract the key points from the top result." Real, current content back means it works.
Common problems
unauthorized— wrong API token.- Credit/quota errors — advanced tools consume credits; check your balance or stick to free-tier tools.
- No tools in client — Node 18+ for
npx; see Cursor MCP not working.
Going further
Bright Data pairs with Sequential Thinking and a filesystem server for research-and-capture. Browse the search category or curated loadouts. Scrape responsibly and within sites' terms; keep your token private — see MCP security best practices.