Repolode
Submit
Skyvern-AI

Skyvern-AI/skyvern

Open-source browser automation that uses LLMs and computer vision to run web workflows on any site, extending Playwright with AI commands.

Open-source alternative to:UiPathZapier

View on GitHubWebsite
Stars
23,233
Forks
2,202
Open issues
288
Last commit
17 hours ago

What is it?

Skyvern automates browser-based workflows by letting vision LLMs read and interact with web pages instead of relying on hand-written XPath selectors or DOM scripts. Because it doesn't depend on fixed selectors, it can work on sites it has never seen before and is less likely to break when page layouts change.

For developers, Skyvern provides a Python and TypeScript SDK that extends Playwright with natural-language commands such as page.act, page.extract and page.validate, and lets standard Playwright calls fall back to AI when a selector fails. For non-technical users, a no-code workflow builder chains tasks together with blocks for data extraction, validation, loops, file parsing, HTTP requests and custom code.

Skyvern can run locally via pip or Docker Compose, using SQLite or Postgres, or through the managed Skyvern Cloud. It works with OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Gemini, Ollama and OpenAI-compatible endpoints, and supports Bitwarden and 1Password-style credential login, 2FA and MCP.

Key features

  • 01Vision LLM-based navigation that doesn't depend on fixed XPath or CSS selectors
  • 02Playwright-compatible SDK with natural-language act, extract, validate and prompt commands
  • 03No-code workflow builder with loops, data extraction, file parsing and HTTP request blocks
  • 04Live browser viewport streaming for watching and intervening in runs
  • 05Structured data extraction driven by a JSON schema
  • 06Credential integrations with Bitwarden and a custom HTTP credential service, plus 2FA (TOTP) support

Use cases

  • Downloading invoices from many different vendor websites on a schedule
  • Filling out contact forms or registration forms on government or business sites
  • Extracting structured data from pages that lack an API
  • Running RPA-style tasks such as procurement or quote retrieval through a browser

Quick start

  1. 1

    Install Skyvern with pip

    pip install "skyvern[all]"
  2. 2

    Start the local server and UI with quickstart

    skyvern quickstart
  3. 3

    Open http://localhost:8080 in your browser to run tasks

Things to know

  • Needs an LLM provider API key configured to run tasks.
  • Licensed under AGPL-3.0, a copyleft license that requires sharing source when offering it as a network service.
  • Windows setup additionally requires Rust and the C++ build tools.

Topics

#browser-automation#llm#computer-vision#playwright#rpa#workflow#web-scraping#python#ai#api

FAQ

What is skyvern?
Open-source browser automation that uses LLMs and computer vision to run web workflows on any site, extending Playwright with AI commands. Skyvern automates browser-based workflows by letting vision LLMs read and interact with web pages instead of relying on hand-written XPath selectors or DOM scripts. Because it doesn't depend on fixed selectors, it can work on sites it has never seen before and is less likely to break when page layouts change.
Is skyvern free to use? What license does it have?
skyvern is open source under the AGPL-3.0 license. Check the license terms before using it in commercial products.
Is skyvern still maintained?
Actively maintained. The last commit to skyvern was on Oct 20, 2026.
What can I use skyvern for?
Downloading invoices from many different vendor websites on a schedule Filling out contact forms or registration forms on government or business sites Extracting structured data from pages that lack an API Running RPA-style tasks such as procurement or quote retrieval through a browser
How do I install skyvern?
Install Skyvern with pip: pip install "skyvern[all]" Start the local server and UI with quickstart: skyvern quickstart Open http://localhost:8080 in your browser to run tasks
Where can I get skyvern?
The source code and installation instructions are on GitHub: https://github.com/Skyvern-AI/skyvern

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects