vibium  by VibiumDev

Browser automation infrastructure for AI agents

Created 1 month ago
2,260 stars

Top 19.8% on SourcePulse

GitHubView on GitHub
Project Summary

Browser automation infrastructure designed for AI agents and humans, Vibium simplifies browser control by providing a single, self-contained binary that manages browser lifecycle, WebDriver BiDi protocol, and exposes an MCP server. This allows AI agents like Claude Code to drive a browser with zero setup, while also serving test automation and other browser-driven tasks.

How It Works

Vibium employs a multi-component architecture centered around a single Go binary called "Clicker." This binary acts as a browser manager, a WebSocket server for the WebDriver BiDi protocol, and an MCP server accessible via stdio. An accompanying JS/TS client, installable via npm, provides an API for developers to interact with the Clicker. The design goal is an "invisible" binary, abstracting away browser management complexities for users, especially AI agents that communicate via the MCP protocol.

Quick Start & Requirements

  • Primary Install: npm install vibium (installs Clicker binary and downloads Chrome/chromedriver automatically). For Claude Code integration: claude mcp add vibium -- npx -y vibium.
  • Prerequisites: Node.js for the JS/TS client. Chrome browser is downloaded by default; can be skipped with VIBIUM_SKIP_BROWSER_DOWNLOAD=1.
  • Resource Footprint: The Clicker binary is approximately 10MB. Setup is designed to be rapid, aiming for "zero to hello world in 5 minutes."
  • Links: Getting Started Tutorial, Quick Reference, V2-ROADMAP.md.

Highlighted Details

  • Single Go binary (~10MB) manages browser lifecycle, BiDi proxy, MCP server, auto-wait, and viewport screenshots.
  • Supports the WebDriver BiDi protocol for advanced browser interaction.
  • Direct MCP server integration simplifies control for AI agents.
  • JS/TS client offers both asynchronous and synchronous APIs.
  • Automatic download and management of Chrome and chromedriver.

Maintenance & Community

Vibium is actively developed, with recent updates focusing on core features like the MCP server and sync API. Version 1 prioritizes the core loop of browser control via MCP and the JS client. The roadmap for V2 includes expanding client support to Python and Java, and introducing features like a memory/navigation layer (Cortex) and AI-powered locators. No community links (Discord/Slack) were specified in the provided README.

Licensing & Compatibility

Vibium is licensed under the Apache 2.0 license. This license is permissive and generally compatible with commercial use and linking within closed-source projects.

Limitations & Caveats

The current V1 release focuses on core browser control via the MCP protocol and JS client. Support for other languages like Python and Java is planned for V2. No specific limitations, alpha status, or known bugs were detailed in the provided README excerpt.

Health Check
Last Commit

6 days ago

Responsiveness

Inactive

Pull Requests (30d)
8
Issues (30d)
40
Star History
2,196 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.