Stagehand

Stagehand is an open source SDK for browser agents. It uses AI to make browser automation more resilient, readable, and production-ready, and it is described as a browser agent framework for writing automations with natural language instructions rather than brittle CSS selectors.

Its core primitives are act, extract, observe, and agent. act performs browser actions from a plain-English instruction such as clicking, filling, navigating, and scrolling. extract pulls structured data from any page with Zod schema validation. observe surfaces what is actionable on a page before an action is taken. agent runs multi-step workflows autonomously when end-to-end execution is needed. The page also says scripts can survive real-world page changes and page redesigns, and that the framework gives both deterministic control and the flexibility of agents.

Stagehand is available in TypeScript and Python. It works locally out of the box with any Chromium browser, and it can be connected to Browserbase’s cloud browsers for production use without code changes. The Browserbase integration is described as providing Agent Identity, action caching, session replay, prompt observability, captcha solving, and zero-infrastructure deployment via Functions. The page also says Stagehand supports major model providers including OpenAI, Anthropic, and Google Gemini via the Vercel AI SDK, and that Model Gateway can access supported models through a single Browserbase API key.

It is fully open source under the MIT license. The page presents it as an AI browser automation framework and says it is intended for agents that need to log in, navigate, extract, and take action on real web interfaces, especially when there is no API or when the site changes without warning.