</>MCP Agents Market
Skill

vibium

by VibiumDev2.9kGoUpdated 2026-09-04

The verification layer for coding agents

Claude CodeGemini CLI

Vibium is a browser automation agent skill that enables AI coding agents to interact with web pages, verify their work, and test web applications. It provides a lightweight verification layer through simple CLI commands for navigation, element interaction, screenshot capture, and form filling. Available as both an installable skill for agent frameworks and an MCP server for structured tool use, Vibium requires zero configuration and automatically downloads Chrome. Built on WebDriver BiDi standards, it offers both synchronous and asynchronous APIs across JavaScript, Python, and Java ecosystems.

Key Features

AI-native CLI skill interface with semantic element mapping and references (@e1, @e2) for easy agent interaction
Zero-configuration setup with automatic Chrome browser download and visible-by-default operation
MCP server support for structured tool use with Claude, Gemini, and other AI assistants
Semantic element finding without CSS selectors (by text, label, placeholder, ARIA role)
Session recording with annotated screenshots and diff mapping to track page changes
Multi-language client libraries (JavaScript/TypeScript async and sync, Python async and sync, Java)
WebDriver BiDi standards-based architecture with ~10MB single binary and no runtime dependencies
Firefox support with video recording capability for sessions

Use Cases

  • 01Enabling coding agents to verify web application functionality by navigating pages and capturing screenshots
  • 02Automated form testing where agents fill inputs, select options, and submit forms to validate behavior
  • 03Visual regression testing with annotated screenshots showing interactive element labels
  • 04End-to-end workflow verification where agents navigate multi-step processes and confirm success messages
  • 05Web scraping and data collection tasks requiring JavaScript execution and dynamic page interaction
  • 06Accessibility auditing by finding elements through ARIA roles and semantic labels

Related Skills

View more

vibium — FAQ

What is Vibium?+

Vibium is a browser automation agent skill that gives AI coding agents the ability to interact with web pages through CLI commands, MCP server tools, or programmatic client libraries. It provides navigation, element interaction, screenshot capture, and verification capabilities designed specifically for AI agents to check their work.

How do I install Vibium as an agent skill?+

Install Vibium globally with 'npm install -g vibium', then add it as a skill using 'npx skills add https://github.com/VibiumDev/vibium --skill vibe-check'. This downloads Chrome automatically and installs the skill to your project's .agents/skills directory.

Which AI clients work with Vibium?+

Vibium works with any agent framework that supports CLI skills or MCP servers, including Claude Code, Gemini CLI, and other AI coding assistants. It can be used as a standalone MCP server with Claude Desktop and similar clients that support the Model Context Protocol.

Does Vibium require API keys or browser installation?+

No API keys are required. Vibium automatically downloads Chrome during installation, requiring zero manual browser setup or configuration.

Is Vibium free to use?+

Yes, Vibium is open source and licensed under Apache 2.0, making it free for both personal and commercial use.

Can I use Vibium in my JavaScript, Python, or Java code?+

Yes, Vibium provides client libraries for JavaScript/TypeScript (npm), Python (pip), and Java (Maven/Gradle) with both synchronous and asynchronous APIs for programmatic browser control.

How do I install vibium?+

Open the source repository on GitHub and follow its README. vibium is a skill — MCP Agents Market links you directly to the official repo.

Is vibium free?+

vibium is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches