What Is Vibium? AI-Native Browser Automation Explained

Vibium is an AI-native browser automation tool. It's built on the modern WebDriver BiDi protocol, ships as a single Go binary, downloads its own Chrome, and includes a built-in MCP server — so both your test scripts and AI agents can drive a real browser with one tool. Vibium was created by Jason Huggins, co-creator of Selenium (2004) and Appium (2012), and is free to install via pip install vibium (Python) or npm install vibium (Node), with official Python and JavaScript/TypeScript clients. Unlike classic WebDriver setups, there's no ChromeDriver to match and no Selenium Grid to run — the binary manages the browser, auto-waits for elements (actionability), and exposes the same commands to humans and LLM agents. In short, Vibium is a lean, BiDi-first automation tool designed for the era of AI coding agents, from the person who started Selenium 21 years ago.

What is Vibium?

Vibium is a browser automation framework for testing and AI agents. Instead of bolting onto older protocols, it speaks WebDriver BiDi — the bidirectional successor to classic WebDriver — which gives it fast, event-driven control of the browser. You install one package, and Vibium handles the browser binary for you.

It was created by Jason Huggins, the co-creator of Selenium and Appium. In his words: "I started the Selenium project 21 years ago. Vibium is what I'd build if I started over today."

How Vibium works

Three design choices define Vibium:

• Single Go binary. No tangled dependency tree. The client library talks to a compact Go process that manages the browser.
• WebDriver BiDi under the hood. Bidirectional events mean Vibium can observe and react to the page, not just fire commands at it — which powers auto-waiting (actionability) and richer data collection.
• Built-in MCP server. Vibium speaks the Model Context Protocol out of the box, so AI agents (Claude Code, Cursor, and others) can use the browser as a tool with zero glue code.

Your first Vibium script

Install the Python client:

pip install vibium

Then launch a browser, visit a page, and take a screenshot:

from vibium import browser_sync as browser
 
vibe = browser.launch()
vibe.go("https://example.com")
 
png = vibe.screenshot()
with open("screenshot.png", "wb") as f:
    f.write(png)
 
vibe.quit()

That's a complete, runnable automation. From here you can find elements, type, click, and build full flows.

How is Vibium different from Selenium and Playwright?

Vibium targets the AI-agent era: BiDi-first, single-binary, MCP built in. Mature tools like Playwright and Selenium have huge ecosystems and multi-language support, while Vibium trades breadth for a lean, AI-native core. See the full Vibium vs Playwright comparison.

Where to go next

• Take a screenshot in Vibium
• Automate a login flow
• Vibium vs Playwright