Skyvern
Open-source AI agent that automates browser workflows using computer vision, not brittle scripts
Screenshots
At a glance
What it does
Open-source AI agent that automates browser workflows with vision, forms, portals, logins, no brittle scripts.
Detailed overview
Skyvern is the open-source answer to Operator-style automation for real back-office RPA. It takes a screenshot, uses a vision-LLM to identify elements, and clicks, fills and navigates to complete multi-step web tasks on sites it has never seen, adapting when layouts change. That makes it ideal for form filling, portal logins, invoice downloads and procurement on legacy or no-API web workflows. It hit 85.85% on the WebVoyager benchmark with its 2.0 release. 21,600+ GitHub stars, YC-backed with $2.7M raised, 30,000+ users as of mid-2026, with Python and TypeScript SDK clients. Note: it ships a CAPTCHA-solver and operates on third-party sites, use it within each site's acceptable-use terms.
Key features
- Vision-LLM identifies page elements
- Completes multi-step web tasks
- Adapts when layouts change
- Python and TypeScript SDKs
Who it's for
Best suited for developers looking for ai agents tools. Open-source AI agent that automates browser workflows using computer vision, not brittle scripts
Tags
More AI Agents tools
See allComments0
Sign in to join the conversation.
No comments yet. Be the first to share your thoughts.