Hyperbrowser Examples

Official examples, Jupyter notebooks, coding-agent skills, and HyperAgent task templates for Hyperbrowser — cloud browsers and web infrastructure for AI agents and automation.

Each project is self-contained. Clone the repo, open a directory, add your API key, and run it.

Table of contents

• About Hyperbrowser
• What's in this repository
• Prerequisites
• Quick start
• Find an example
• Example apps
• Cookbook
• Skills
• Task templates
• Repository layout
• Related projects
• Contributing
• Support
• License

About Hyperbrowser

Hyperbrowser is browser infrastructure for AI agents and apps: managed cloud browser sessions, scrape / crawl / extract APIs, computer-use agents, and sandboxes.

Use it when you need a real browser in the cloud — JavaScript-rendered pages, stealth and proxies, CAPTCHA handling, Playwright or Puppeteer over CDP, or an agent that can click through a site.

• Product: hyperbrowser.ai
• Dashboard / API keys: app.hyperbrowser.ai
• Docs: hyperbrowser.ai/docs
• Node SDK: @hyperbrowser/sdk
• Python SDK: hyperbrowser

What's in this repository

Area	Path	What it is
Example apps	Top-level directories	Standalone CLIs, bots, and small apps in TypeScript or Python
Cookbook	cookbook/	Jupyter notebooks for scrape, extract, Browser Use, and MCP patterns
Skills	skills/	Prompts you can drop into a coding agent (/web, /design, /learn, …)
Task templates	task-templates/	Copy-paste HyperAgent scripts by industry

There is no root package.json. Install and run inside the directory you care about.

Prerequisites

• A Hyperbrowser API key
• TypeScript examples: Node.js 18 or newer, plus npm
• Python examples: Python 3.10+ (several Streamlit apps specify 3.13), plus pip or uv
• Many examples also need an OpenAI API key. A few use Anthropic, Groq, Slack, Serper, or ElevenLabs — each project's README lists the keys it needs

Never commit .env files. The repo gitignores them.

Quick start

git clone https://github.com/hyperbrowserai/examples.git
cd examples/company-researcher
npm install

Create a .env in that directory:

HYPERBROWSER_API_KEY=your_api_key_here

Run the example:

npx ts-node company-researcher.ts

Enter a company name and a research topic. The script searches the web and extracts structured findings through Hyperbrowser.

Python path (notebooks):

cd cookbook
pip install -r requirements.txt
jupyter notebook

Add HYPERBROWSER_API_KEY (and any other keys the notebook asks for) to cookbook/.env.

Find an example

If you want to…	Start here
Scrape a page and chat with it	ChatWithWebsite-Scrape
Research a company or competitor	company-researcher, competitor-analyzer-bot
Turn sites into LLM / RAG data	site2rag, ragzip, llm-crawl, hyper-train
Monitor the web and alert Slack	research-bot, hb-intern-bot, down-monitor
Audit UX, SEO, or security headers	SEO-Analyzer, hb-headers, dark-pattern-finder
Drive a browser with an AI agent	task-templates, cookbook Browser Use notebooks
Teach a coding agent a live site	skills/web-skill.md, skills/design-skill.md
Compare computer-use models	browser-agent-benchmark

Example apps

Each folder has its own README with install steps, environment variables, and commands.

Research and competitive intelligence

Project	Description	Stack
company-researcher	Research any company on a topic and return structured findings	TypeScript
competitor-analyzer-bot	Scrape two competitor sites and generate a comparison report	TypeScript
research-bot	Watch competitor pages, detect real changes, and send founder-focused Slack alerts	TypeScript
hb-intern-bot	Scheduled research intern: scrape HN, Reddit, Product Hunt, and blogs into digests	TypeScript
hb-predict	Detect emerging tech signals from HN and Reddit and generate predictions	TypeScript
Extract-github-analyzer	Analyze a GitHub profile: languages, frameworks, and top repositories	TypeScript
github-chatbot	Scrape a GitHub repo and answer questions about it	TypeScript
Trend-summary	Take today's top Hacker News post and summarize related Reddit discussion	TypeScript

Scraping, crawl, and LLM data

Project	Description	Stack
ChatWithWebsite-Scrape	Scrape any URL and chat with the page in the terminal	TypeScript
llm-crawl	Crawl with the official Crawl API, then process pages with an LLM	TypeScript
site2rag	Scrape, strip boilerplate, and emit token-budgeted RAG chunks	TypeScript
site2prompt	Scrape and clean sites into LLM training / prompt datasets	TypeScript
ragzip	Build citation-tagged context packs from scraped pages	TypeScript
hyper-train	Scrape URLs into JSONL / Markdown training sets, with optional embeddings and QA pairs	TypeScript
dataset-assmbler	Search the web, extract text, dedupe, and split train/eval JSONL or CSV	TypeScript
Internet-zip	Compress a live page into a semantic .kzip.json knowledge shard	TypeScript
oss-web-extractor	Extract structured data with Hyperbrowser plus a local open-source model	TypeScript
o3-pro-extractor	Scrape a page to Markdown, then extract structured JSON with OpenAI	TypeScript
meta-scraper	Extract and analyze title, description, Open Graph, and Twitter Card tags	TypeScript
site-graph	Crawl a domain and print a site map, orphan pages, and heaviest pages	TypeScript
resource-summary	Summarize the resources a page loads	TypeScript
tweet-fetcher	Extract recent tweets and followers for a handle using a persistent browser profile	Python

Monitoring, news, and alerts

Project	Description	Stack
down-monitor	Real-browser uptime checks (not HTTP pings)	TypeScript
down-detector-bot	Watch AWS, GCP, Azure, and Cloudflare status pages; Slack on change	TypeScript
hb-changelog-tracker	Monitor AI / research blogs and post changelog-style Slack summaries	TypeScript
agi-newsletter	Scrape AI lab blogs, draft a newsletter, and email subscribers	TypeScript
crypto-news-bot	Aggregate CoinDesk, Decrypt, and Cointelegraph into Slack digests	TypeScript
vibe-posting-bot	Detect new tech stories and draft human-like Typefully posts	TypeScript
link-sniper-bot	Find external links on a page and classify them as working, broken, or blocked	TypeScript

Commerce, leads, and local search

Project	Description	Stack
product-search	Extract a product, find Google Shopping alternatives, and refresh prices on a schedule	TypeScript
real-estate-finder	Search US rental and for-sale listings across major real-estate sites	TypeScript
Maps-lead-finder	Find business leads on Google Maps (name, address, contact) via HyperAgent	TypeScript
gift-researcher	Research gift ideas from live web discussions	TypeScript
scam-scanner-bot	Scan a storefront in a real browser and flag fraud patterns	TypeScript

Site quality, privacy, and UX

Project	Description	Stack
SEO-Analyzer	AI SEO audit: titles, meta, headings, content, images, links, and technical issues	TypeScript
hb-headers	Follow redirects and report CORS, CSP, HSTS, cookies, and frame options	TypeScript
hb-ui-bot	Screenshot a page and review layout, type, color, and accessibility with vision	TypeScript
dark-pattern-finder	Scan for deceptive UX (fake scarcity, hidden fees, obstruction)	TypeScript
dataflow-tree	Map PII collection, forms, and third-party trackers	TypeScript
deep-form	Reverse-engineer form fields, validation, and submission flows	TypeScript
CUA-CTA-Validator	Use OpenAI Computer Use to find the hero CTA and critique accessibility / SEO	TypeScript

Agents, apps, and evaluation

Project	Description	Stack
chat-with/chat-with-youtube	Streamlit app: extract a YouTube transcript and chat with the video	Python
article-tts	Streamlit app: fetch an article and convert it to speech	Python
changelog-builder	Streamlit app: compare two GitHub refs and generate a changelog	Python
agents/budget-travel-agent	Streamlit app: search Google Travel Explore and parse results with vision	Python
browser-agent-benchmark	Fair harness: run multiple models on the same real-browser tasks and score them	TypeScript

Cookbook

Jupyter notebooks in cookbook/. Install cookbook/requirements.txt, set HYPERBROWSER_API_KEY (and usually OPENAI_API_KEY) in .env, then open any notebook.

Notebook	Topic
company-researcher.ipynb	Company research
docs-qna.ipynb	Question answering over documentation
documentation-based-coding-agent.ipynb	Coding agent grounded in live docs
youtube_video_chat.ipynb	Chat with a YouTube transcript
hacker-news-summarizer.ipynb	Summarize Hacker News
news-analyst.ipynb	News analysis
twitter-profile-analyzer.ipynb	Twitter / X profile analysis
social-media-post-finder.ipynb	Find social posts
shopping-assistant.ipynb	Shopping assistant
shopping-agents-with-vision.ipynb	Vision-based shopping agents
review-analyzer.ipynb	Review analysis
movie-review-researcher.ipynb	Movie review research
apartment-finder.ipynb	Apartment search
flight-ticket-search.ipynb	Flight search
concert-ticket-finder.ipynb	Concert tickets
local-events-finder.ipynb	Local events
steam-special-suggestor.ipynb	Steam specials
doordash-location-recommender.ipynb	Delivery / restaurant recommendations
menu-recommendations.ipynb	Menu recommendations
ingredients-based-recipe-suggestor.ipynb	Recipes from ingredients
changelog-builder.ipynb	Changelog generation
wiki-racer.ipynb	Path finding between Wikipedia articles
next-chess-move.ipynb	Next chess move from a live board
code-solver.ipynb	Solve coding problems in the browser
code-solver-browser-use.ipynb	Same pattern with Browser Use
browser-use-hybrid-flows.ipynb	Hybrid Browser Use flows
browser-use-authed-flows.ipynb	Authenticated Browser Use flows
feature-recognition.ipynb	Recognize product features on a page
comic-trip-planner.ipynb	Trip planning
youtube-mcp-server.ipynb	YouTube MCP server
wikipedia-mcp-server.ipynb	Wikipedia MCP server
news-mcp-server.ipynb	News MCP server

Skills

Markdown skills in skills/ for coding agents (Claude Code, Cursor, and similar). They use Hyperbrowser's Fetch API or MCP tools (web_fetch, scrape_webpage) and, for discovery, Serper.

Skill	Command	What it does
web-skill.md	/web <url>	Map a live site into a web.md (pages, actions, flows, interactive elements)
design-skill.md	/design <url>	Extract colors, type, spacing, and components into DESIGN.md
harness-skill.md	/harness	Run the agent against your repo and write a CLAUDE.md from real failures
learn.md	/learn <topic>	Discover official docs and generate a skill for a library or API
learn-batch.md	/learn batch <topics…>	Same as /learn, for several topics at once
skill-tree.md	/skill-tree <topic-or-url>	Build a skill graph from documentation

Copy a skill into your agent's skills directory (for example ~/.claude/skills/ or .claude/skills/) and invoke the command. You need HYPERBROWSER_API_KEY; /learn and /skill-tree also need a Serper key.

Task templates

Ready-to-run HyperAgent scripts in task-templates/. Each file is a focused automation with a Zod schema. Copy one, set HYPERBROWSER_API_KEY, and execute it.

Category	Templates
AI pipelines	News sentiment, product-review analysis, Amazon and TikTok shopping recommenders
API servers	Craigslist, Hacker News, Product Hunt, Wikipedia, multi-site
Booking	Google Flights, Kayak, hotels, OpenTable, Calendly
Concepts	Tables, infinite scroll, SEO content structure
Data extraction	Zillow, Airbnb, X / Twitter
E-commerce	Multi-site price comparison, coupons, eBay auctions, Etsy shops
Finance	Stock quotes, earnings calendars, FX rates, crypto prices
Form filling	Basic forms, Google Forms, multi-step onboarding
Job search	Indeed
Monitoring	Changelog tracker, stock availability
Research	Competitor features, GitHub trending, startup launches, App Store
Social	YouTube, TikTok, Threads, Instagram
Utils	Session video recording helpers

Repository layout

examples/
├── cookbook/              # Jupyter notebooks
├── skills/                # Coding-agent skills
├── task-templates/        # HyperAgent scripts by category
├── agents/                # Multi-step agent apps
├── chat-with/             # Chat-with-content apps
├── <example>/             # One standalone project per directory
├── LICENSE                # MIT
└── README.md

Typical TypeScript example:

example-name/
├── package.json
├── tsconfig.json
├── *.ts
└── README.md

Typical Python example:

example-name/
├── pyproject.toml
├── *.py
└── README.md

Related projects

Project	Description
hyperbrowserai/node-sdk	Official Node.js SDK
hyperbrowserai/python-sdk	Official Python SDK
hyperbrowserai/HyperAgent	Open-source Playwright agent framework
hyperbrowserai/mcp	Model Context Protocol server
hyperbrowserai/n8n-node	n8n integration
hyperbrowserai/hyperbrowser-app-examples	Full Hyperbrowser-powered web apps

Contributing

New examples are welcome.

1. Fork the repository and create a branch.
2. Add a self-contained directory (or a notebook under cookbook/, a skill under skills/, a template under task-templates/).
3. Include a README that covers purpose, required keys, install, and a command that actually runs.
4. Use @hyperbrowser/sdk or hyperbrowser — do not reimplement the HTTP API.
5. Load secrets from the environment. Do not commit .env, API keys, node_modules, .venv, or generated datasets.
6. Keep the example focused. Prefer a small, runnable demo over a framework.
7. Open a pull request against main.

Please follow the site's terms of service and robots rules for any URL an example hits by default.

Support

• Documentation: hyperbrowser.ai/docs
• Discord: discord.gg/zsYzsgVRjh
• Email: info@hyperbrowser.ai
• Issues: github.com/hyperbrowserai/examples/issues

Running an example uses your Hyperbrowser (and any LLM) quota. Failed jobs and long crawls still consume credits — start with a single URL and a short timeout while you learn the API.

License

This repository is licensed under the MIT License.