Bright Data

Bright Data Hackathon Kit

Scrape any website. Get any data. Build something amazing.
Your one-stop resource for building with Bright Data at hackathons.

Quick Start • MCP & Agents • Products • Examples • Showcase • Help

Hi, I'm Rafael

I'm a DevRel Engineer at Bright Data. I'm here to help you build!

What is Bright Data?

Bright Data gives you tools to collect data from any website without getting blocked. Our APIs handle CAPTCHAs, anti-bot systems, browser fingerprinting, and IP rotation — so you can focus on building your hack, not fighting websites.

Full access to all Bright Data products.

Get Started in 60 Seconds

1. Get your API key

Go to brightdata.com/cp/setting/users and copy your API key.

2. Know your zones

A zone is a configured instance of a Bright Data product. Think of it as a named access point — each product (Web Unlocker, SERP API, Scraping Browser) gets its own zone with its own credentials and settings.

Find your zones at brightdata.com/cp/zones
Each zone has a name (e.g., web_unlocker1, serp_api1) that you pass in API calls
Scraping Browser zones also have a username and password (used for WebSocket auth)
When you sign up, default zones are created for you — check your control panel

3. Set your environment variables

export BRIGHTDATA_API_KEY="your_api_key_here"

# Zone names — find yours at brightdata.com/cp/zones
export BRIGHTDATA_UNLOCKER_ZONE="web_unlocker1"     # Web Unlocker zone
export BRIGHTDATA_SERP_ZONE="serp_api1"              # SERP API zone

4. Scrape something

Python:

import requests

response = requests.post(
    "https://api.brightdata.com/request",
    headers={
        "Authorization": f"Bearer {YOUR_API_KEY}",
        "Content-Type": "application/json",
    },
    json={
        "zone": "web_unlocker1",
        "url": "https://www.example.com",
        "format": "raw",
    },
)
print(response.text)  # HTML of the page

JavaScript:

const response = await fetch("https://api.brightdata.com/request", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${YOUR_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    zone: "web_unlocker1",
    url: "https://www.example.com",
    format: "raw",
  }),
});
console.log(await response.text()); // HTML of the page

cURL:

curl -X POST "https://api.brightdata.com/request" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"zone": "web_unlocker1", "url": "https://www.example.com", "format": "raw"}'

That's it. You just scraped a website.

MCP & AI Agents

The fastest way to build an AI agent with web access. MCP (Model Context Protocol) lets your agent search the web, scrape pages, and extract structured data — without you writing API calls.

Zero-Setup Option

Add this URL to any MCP-compatible client (Claude Desktop, Cursor, Windsurf):

https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN

Claude Desktop Setup

{
  "mcpServers": {
    "Bright Data": {
      "command": "npx",
      "args": ["@brightdata/mcp"],
      "env": {
        "API_TOKEN": "YOUR_API_TOKEN"
      }
    }
  }
}

Build an Agent in 10 Lines

from llama_index.tools.brightdata import BrightDataToolSpec
from llama_index.llms.openai import OpenAI
from llama_index.core.agent.workflow import FunctionAgent

agent = FunctionAgent(
    tools=BrightDataToolSpec(api_key=API_KEY, zone="mcp_unlocker").to_tool_list(),
    llm=OpenAI(model="gpt-4o-mini"),
)
response = await agent.run("Find the top AI trends this week and summarize them")

The agent autonomously decides which tool to use:

Web search → search_engine
Scrape a page → scrape_as_markdown
Amazon product → web_data_amazon_product
LinkedIn profile → web_data_linkedin_person_profile
60+ more tools available

Free tier: 5,000 requests/month — more than enough for a hackathon.

Full MCP Guide → | Agentic Workflow Patterns →

Products

Product	What It Does	Best For
MCP Server	Give AI agents real-time web access	Agentic workflows, LLM tools
Web Scraper API	Returns structured JSON from popular sites	Amazon, LinkedIn, Instagram data
SERP API	Search engine results without blocks	Google search, market research
Scraping Browser	Remote browser via Puppeteer/Playwright	JS-heavy sites, interactions, screenshots
Web Unlocker	Fetches any page's HTML without blocks	Any website, raw HTML

Not sure which to use? Check the decision tree →

Code Examples

Ready-to-run scripts in Python and JavaScript. Copy, paste, run.

Python

Example	What It Does
`01_web_scraper_api.py`	Get structured JSON from supported sites
`02_serp_api.py`	Search Google and get parsed results
`03_scraping_browser.py`	Control a remote browser with Playwright
`04_web_unlocker.py`	Fetch any page's HTML
`05_mcp_agent.py`	Build an AI agent with web access (MCP)

cd examples/python
pip install -r requirements.txt    # Installs all deps (including MCP agent)
export BRIGHTDATA_API_KEY="your_key"
export BRIGHTDATA_UNLOCKER_ZONE="web_unlocker1"
python 04_web_unlocker.py          # Try the simplest example first

# For the MCP agent (05), you also need:
export OPENAI_API_KEY="your_openai_key"
python 05_mcp_agent.py

JavaScript

Example	What It Does
`01_web_scraper_api.js`	Get structured JSON from supported sites
`02_serp_api.js`	Search Google and get parsed results
`03_scraping_browser.js`	Control a remote browser with Puppeteer
`04_web_unlocker.js`	Fetch any page's HTML
`05_mcp_agent.js`	Connect to Bright Data MCP server

cd examples/javascript
npm install
export BRIGHTDATA_API_KEY="your_key"
export BRIGHTDATA_UNLOCKER_ZONE="web_unlocker1"
node 04_web_unlocker.js             # Try the simplest example first

Python SDK (Even Simpler)

pip install brightdata

from brightdata import BrightDataClient

async with BrightDataClient() as client:
    result = await client.scrape_url("https://example.com")
    search = await client.search.google(query="python scraping")

Starter Template

A fork-ready Next.js app with Bright Data already wired up:

cd starter-template
npm install
cp .env.example .env     # Add your API key
npm run dev               # Open http://localhost:3000

Enter a URL → get scraped HTML. Build on top of it.

See the starter template →

Built with Bright Data

Real projects from our team. Use these for inspiration, reference code, or fork them as your starting point.

Project	What It Does	Stars	Stack
PepoLeHub	Natural language people search engine	320+	Next.js, LangGraph, Bright Data
MCP_ADK	Web search agent with Google ADK	46	Python, Gemini, Bright Data MCP
FactFlux	Multi-agent fact-checking system	41	Agno, Gemini, Bright Data
Unified Search	Smart query routing agent	18	LangGraph, Bright Data MCP
The Missing Link in AI	MCP-powered research agent	-	Node.js, LangChain, Bright Data MCP
Internet Mood Radar	Real-time global sentiment tracker	-	Next.js, OpenAI, Bright Data
Rent Hunter	Multi-agent apartment search	10	LangGraph, Streamlit, Bright Data
Geo Chat	Location-based search with maps	-	Express.js, LangGraph, MapLibre
EnrichIt	AI spreadsheet data enrichment	3	React, FastAPI, LangGraph
AI Travel Planner	Travel assistant with real-time data	-	FastAPI, LangGraph, WebSocket

See all projects with details →

Project Ideas

Need inspiration? We've got 20+ ideas organized by hackathon track:

Healthcare — Drug price tracker, clinical trial monitor
Sustainability — Greenwashing detector, eco-product finder
AI/ML — Training data pipeline, RAG knowledge base builder
Fintech — Price comparison engine, alternative trading data
Education — Scholarship aggregator, course comparison tool

See all project ideas →

Quick Reference

MCP Guide — Set up MCP and build AI agents with web access
Agentic Workflows — Architecture patterns for AI + web data
Product Decision Tree — Which product should I use?
API Cheat Sheet — Endpoints, auth, response formats at a glance
Troubleshooting — Common issues and fixes
Project Showcase — Real projects built with Bright Data

Official Resources

Need Help?

Open an issue on this repo
Reach out on LinkedIn — happy to help!

Good luck! Build something amazing.

brightdata-hack-pack

Links

README