Skyvern

Skyvern automates browser tasks for teams handling forms, vendor portals and repeated data collection. It uses vision models to identify page elements, helping workflows adapt when website layouts change.

• Workflow building: Create workflows through natural-language instructions, browser recordings or a visual builder, or use the Playwright-compatible SDK from code.
• Forms and logins: Fill forms and work through authentication flows, including two-factor authentication, TOTP and CAPTCHAs, with Bitwarden credential support.
• Data and documents: Extract structured results as JSON or CSV, download files, and combine document parsing with validation and HTTP requests.
• Run inspection: Watch the browser viewport live and inspect summaries of the agent's actions and reasoning.
• App connections: Connect workflows to Zapier, Make, n8n, Clay and Workato, or expose browser automation to AI tools through MCP.

Skyvern is open source under AGPL-3.0 and can run on your own server in Docker. It supports local models through Ollama, OpenAI-compatible endpoints, and cloud providers including OpenAI, Anthropic and Gemini; cloud model requests go to the selected provider. Skyvern Cloud is a separate hosted option that can control Chrome on your machine using that browser's existing logins and extensions. A locally running browser in this arrangement is still controlled by the cloud service.