Skip to content

skyvernvsbrowser-use

AGPL-3.0 148 17 21,046
250.9 thousand (month) Feb 01 2024 1.0.29(2026-04-02 14:42:44 ago)
87,251 30 226 MIT
Nov 01 2024 8.9 million (month) 0.12.6(2026-04-02 07:55:13 ago)

Skyvern is an AI-powered browser automation tool that uses large language models (LLMs) and computer vision to interact with websites. Instead of relying on DOM selectors, Skyvern takes screenshots of web pages and uses visual understanding to identify and interact with elements, making it highly resilient to website changes.

Key features include:

  • Vision-based interaction Uses screenshots and computer vision (multimodal LLMs) to understand page layout and identify interactive elements visually, rather than through DOM inspection alone.
  • No selectors needed Describe tasks in natural language and Skyvern figures out what to click, type, and navigate without CSS selectors or XPath.
  • Complex workflow automation Can handle multi-step workflows like form filling, navigation through menus, file uploads, and multi-page processes.
  • Self-correcting When actions fail, Skyvern can analyze the resulting page state and adjust its approach, recovering from errors autonomously.
  • API-first design Provides a REST API for triggering and managing automation tasks programmatically.
  • Open source with cloud option Core engine is open source and can be self-hosted. Also available as a managed cloud service.

Skyvern is particularly effective for automating tasks on websites with complex or dynamic UIs where traditional selector-based automation breaks frequently. It achieved 85.85% accuracy on the WebVoyager benchmark.

Browser-use is a Python library that enables AI agents to control web browsers using natural language instructions. It connects large language models (LLMs) to browser automation, allowing you to describe what you want done in plain English instead of writing explicit selectors and interaction code.

Key features include:

  • Natural language browser control Describe tasks like "go to Amazon and find the cheapest laptop under $500" and the AI agent will navigate, interact with elements, and extract the requested information.
  • Multi-step task execution Can handle complex workflows that require multiple pages, form filling, clicking, scrolling, and waiting for dynamic content.
  • Vision support Uses screenshot analysis (multimodal LLMs) to understand page layout and find elements visually, not just through DOM inspection.
  • Multiple LLM providers Works with OpenAI, Anthropic Claude, Google Gemini, and other LLM providers.
  • Playwright backend Uses Playwright under the hood for reliable browser automation across Chrome, Firefox, and Safari.
  • Structured output Can return extracted data in structured formats defined by Pydantic models.

Browser-use represents a new paradigm in web scraping where instead of writing brittle selectors, you describe the extraction task and let the AI figure out how to navigate and extract the data. This is especially useful for scraping diverse sites with varying layouts.

Highlights


ai-powerednatural-languageanti-detect
ai-powerednatural-languageasync

Example Use


```python import requests # Skyvern runs as a service - interact via REST API SKYVERN_API = "http://localhost:8000/api/v1" # Create a task with natural language instructions task = requests.post( f"{SKYVERN_API}/tasks", json={ "url": "https://example.com/contact", "navigation_goal": "Fill out the contact form with test data and submit it", "data_extraction_goal": "Extract the confirmation message after submission", "navigation_payload": { "name": "John Doe", "email": "john@example.com", "message": "Hello, this is a test message", }, }, ).json() task_id = task["task_id"] # Check task status result = requests.get(f"{SKYVERN_API}/tasks/{task_id}").json() print(result["status"]) # "completed" print(result["extracted_information"]) # confirmation message ```
```python from browser_use import Agent from langchain_openai import ChatOpenAI import asyncio async def main(): # Create an AI agent with a language model agent = Agent( task="Go to reddit.com/r/webscraping, find the top 5 posts " "from today, and extract their titles and scores", llm=ChatOpenAI(model="gpt-4o"), ) # Run the agent - it navigates and extracts automatically result = await agent.run() print(result) # More complex multi-step task agent = Agent( task="Go to example.com/login, log in with user@test.com " "and password 'test123', then navigate to the dashboard " "and extract all notification messages", llm=ChatOpenAI(model="gpt-4o"), ) result = await agent.run() print(result) asyncio.run(main()) ```

Alternatives / Similar


Was this page helpful?