Bright Data MCP Server – Live Web Access for AI
Bright Data's MCP server gives MCP-capable AI agents access to public web data through hosted or local server options. It supports web search, page scraping, structured web-data tools, browser automation, and configurable tool groups for research, ecommerce, finance, social, travel, and developer package workflows.
Overview
Bright Data MCP Server, also called The Web MCP, connects AI agents to Bright Data's public web access infrastructure through the Model Context Protocol. It is designed for agents that need current public web information, page extraction, search results, structured datasets, and controlled browser sessions without building a separate scraping pipeline.
What the MCP server enables
The server exposes tools for web search, webpage extraction, structured platform data, and browser automation. The default Rapid mode includes web search, Markdown scraping, and AI-ranked discovery tools. Pro mode and tool groups expand access to advanced scraping, structured data feeds, ecommerce and social platform extractors, business and finance datasets, app store and travel data, GEO and LLM visibility tools, code package lookups, and browser-control tools.
Bright Data documents examples such as search_engine, scrape_as_markdown, discover, scrape_as_html, extract, web_data_* platform tools, and scraping_browser_* browser automation tools. Some structured tools require specific URL formats, such as product URLs, profile URLs, listing URLs, or package names.
When to use it
Use Bright Data MCP when an AI agent needs fresh public web context, competitive research, product and price intelligence, social or business profile enrichment, package metadata from npm or PyPI, or browser interaction with dynamic pages. It is also useful for research agents that need search, scrape, and extract operations from one MCP server.
Connection and authentication
Bright Data supports a hosted remote MCP endpoint and a local self-hosted MCP server. The hosted Streamable HTTP endpoint is configured with a Bright Data API token in the query string. The documented local setup runs the npm package with npx and passes the token as the API_TOKEN environment variable. Bright Data also documents advanced local environment variables such as PRO_MODE, GROUPS, TOOLS, rate limits, browser zones, and timeout settings.
Key considerations
Bright Data's free MCP tier is account-level and documented as 5,000 monthly requests from a shared credit pool. Rapid mode is the default; Pro mode and some tool groups can incur usage-based charges. Agents should enable only the tool groups they need to reduce context size and usage. Scraped web content should be treated as untrusted input, validated before use in prompts, and handled carefully to reduce prompt-injection and data-quality risks.
Supported Transports
streamable_http
URL: https://mcp.brightdata.com/mcp?token=${BRIGHTDATA_API_TOKEN}
sse
URL: https://mcp.brightdata.com/sse?token=${BRIGHTDATA_API_TOKEN}
stdio
Command: npx
Args:
@brightdata/mcp
Frequently Asked Questions
- When should an AI agent use Bright Data MCP Server?
- Use it when an agent needs live public web data, search results, webpage extraction, structured ecommerce or social data, package metadata, or browser automation instead of relying on stale model knowledge.
- What does Bright Data MCP add to an AI agent's capabilities?
- It adds MCP tools for search, Markdown or HTML scraping, AI-ranked discovery, structured web-data feeds, browser-control actions, and optional tool groups for ecommerce, social, finance, business, research, app store, travel, GEO, code, and advanced scraping workflows.
- What can the AI agent access or manage through Bright Data MCP?
- The agent can access public web search and page content, structured platform data such as product pages, social profiles, company data, app store entries, travel listings, code package metadata, and browser session operations. Availability depends on the enabled mode, selected tool groups, and valid input formats for each tool.
- How is authentication configured for Bright Data MCP Server?
- Authentication uses a Bright Data API token. The hosted endpoint places the token in the MCP URL query string, while the local npm-based server expects the token in the `API_TOKEN` environment variable. Store the token in a secret manager or environment variable and avoid committing it to client configuration files.
- Which transport should be used for Bright Data MCP Server?
- Use the hosted Streamable HTTP endpoint for the simplest managed setup. Use the documented SSE endpoint only when an MCP client requires SSE compatibility. Use the local stdio server when you want to run the npm package yourself and configure environment variables such as `PRO_MODE`, `GROUPS`, `TOOLS`, and timeout options locally.