Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.agentstack.sh/llms.txt

Use this file to discover all available pages before exploring further.

​
Description

Hyperbrowser enables your agents to interact with the web through powerful browser automation. This tool allows agents to extract content from webpages, crawl entire sites, extract structured data, and perform sophisticated browser automation tasks using multiple AI agent options.

​
Example

Add the Hyperbrowser tool with 
agentstack tools add hyperbrowser
Set up your environment variables: 
HYPERBROWSER_API_KEY=your_api_key_here

​
Features

​
Web Scraping

Extract content from individual webpages in various formats: 
from agentstack._tools.hyperbrowser import scrape_webpage

result = scrape_webpage(
    url="https://example.com",
    use_proxy=True,
    formats=["markdown"]  # Options: markdown, html, links, screenshot
)

​
Website Crawling

Crawl entire websites and collect content from multiple pages: 
from agentstack._tools.hyperbrowser import crawl_website

result = crawl_website(
    starting_url="https://example.com",
    max_pages=10,
    include_pattern=["/blog/*"],
    exclude_pattern=["/admin/*"],
    use_proxy=True
)

​
Structured Data Extraction

Extract data in structured format based on a schema and prompt: 
from agentstack._tools.hyperbrowser import extract_data_from_webpages

result = extract_data_from_webpages(
    urls=["https://example.com/product"],
    schema="{\"type\":\"object\",\"properties\":{\"name\":{\"type\":\"string\"},\"price\":{\"type\":\"string\"}}}",
    prompt="Extract the product name and price from this page",
    system_prompt="You are an expert at extracting product information",
    use_proxy=True
)

​
Browser Use Agent

Run a fast, efficient browser automation agent for explicit instructions: 
from agentstack._tools.hyperbrowser import run_browser_use_agent

result = run_browser_use_agent(
    task="Go to example.com and click on the first link",
    max_steps=10,
    use_vision=False,
    use_vision_for_planner=False,
    use_proxy=True
)

​
Claude Computer Use Agent

Leverage Claude’s advanced reasoning capabilities for complex web tasks: 
from agentstack._tools.hyperbrowser import run_claude_computer_use_agent

result = run_claude_computer_use_agent(
    task="Research the top 3 AI frameworks and summarize their features",
    max_steps=20,
    use_vision=True,
    use_vision_for_planner=True,
    use_proxy=True
)

​
OpenAI Computer Use Agent

Utilize OpenAI’s balanced performance for general-purpose browser automation: 
from agentstack._tools.hyperbrowser import run_openai_cua_agent

result = run_openai_cua_agent(
    task="Find the latest news on AI development",
    max_steps=15,
    use_vision=True,
    use_vision_for_planner=False,
    use_proxy=True
)

​
Available Functions

The Hyperbrowser tool provides the following core functions:
  • scrape_webpage(): Extract content from a single webpage in various formats
  • crawl_website(): Collect content from multiple pages across a website
  • extract_data_from_webpages(): Extract structured data from webpages based on a schema
  • run_browser_use_agent(): Fast, lightweight agent for explicit browser automation tasks
  • run_claude_computer_use_agent(): Claude-powered agent for complex reasoning tasks
  • run_openai_cua_agent(): OpenAI-powered agent for general-purpose browser automation
For detailed function parameters and usage, refer to the function docstrings in your IDE or the Hyperbrowser documentation.