Skyvern
An open-source AI browser automation tool with a no-code builder, a Playwright-compatible SDK, and support for Ollama and cloud models.
Skyvern uses vision models to carry out tasks in a browser, such as filling forms, collecting data, and working through logins. It's for teams automating websites whose layouts change, as well as developers who want AI actions alongside Playwright. Instead of depending on fixed page selectors, it identifies visible elements and decides how to interact with them.
A no-code builder lets nontechnical users assemble workflows, while a Playwright-compatible SDK gives developers a way to add the same capabilities to code. Workflows can join browser tasks with data extraction, validation, file parsing, and HTTP requests. Skyvern can return extracted data in a specified structure. It also downloads files.
The project is open source under AGPL-3.0. It supports locally hosted models through Ollama, as well as models from OpenAI, Anthropic, Gemini, Azure OpenAI, AWS Bedrock, OpenRouter, and OpenAI-compatible endpoints. Skyvern Cloud can control a Chrome browser running on your machine, with access to that browser's existing logins and extensions. For workflows behind accounts, Skyvern supports two-factor authentication and Bitwarden credentials. Zapier, Make.com, and n8n integrations connect browser workflows to other apps.
