Skyvern

Skyvern is an open-source AI agent for browser automation that uses computer vision and LLMs instead of hard-coded selectors. It screenshots each page, has a vision model locate the right elements, and interacts with them, so it can run on unfamiliar sites without custom scripting or selector upkeep. It can be self-hosted for free (AGPL-3.0) or used via a managed cloud with tiered, credit-based pricing. It is popular with developers and YC-backed.

Skyvern is an open-source (AGPL-3.0) AI agent that automates tasks in a web browser. Instead of relying on hard-coded selectors that break when a site changes, it takes a screenshot of the current page, uses a vision-capable LLM to identify the relevant elements and their location, and interacts with them. This visual reasoning approach lets Skyvern operate on websites it has never encountered without custom scripting or selector maintenance. Because it is open source, teams can self-host Skyvern for free and run it inside their own infrastructure, which is attractive for cost control and data sensitivity. Skyvern also offers a managed cloud with tiered plans, adding conveniences like predictable credit limits, concurrency, anti-bot handling, and authentication support. It has become a popular project, with tens of thousands of GitHub stars. Skyvern is a Y Combinator-backed company that raised a $2.7M round to build the project. As with any browser agent, real-world reliability varies by site complexity and anti-bot defenses, so it is best evaluated on the specific workflows you intend to automate.