Vibeleaderboard
← All Intel
Intel / article

Stagehand

Source
browserbase.com
Author
Browserbase
Date

Context

This item is Browserbase's blog index for Stagehand, its open-source framework for browser automation driven by AI models. The index itself is a list of posts, so the useful question is what Stagehand is and what the list shows about where it has been heading.

According to the Stagehand repository, the framework lets developers mix ordinary Playwright-style browser code with plain-language instructions. Three calls do most of the work: observe finds page elements from a description and returns selectors, act performs a described action and adapts when a site's layout changes, and extract returns data matching a schema the developer defines. Because observe hands back selectors, the developer's own code can type in passwords and other sensitive values without those ever reaching the model. It is MIT-licensed and offered in TypeScript, Python and Go.

The posts on the index trace the main pressures on that design. Browserbase explained why it was moving away from depending on Playwright and launched Stagehand v3 in October 2025, then wrote about supporting every language, about the risk of malicious page content steering an AI browser, about caching repeated calls so they do not spend model each time, and, in the index's latest entry from April 2026, about a Model Gateway that removes the need to manage model keys. Since then, Browserbase has released Stagehand v4, which it describes as the SDK for browser agents.

Terms in this piece · Glossary
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
  • LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
Read the source browserbase.com
More from Browserbase
Recommended reads
Comments

Checking sign-in…

Loading comments…