Skip to content

Getting Started

Welcome to vibe-check! This tutorial section will guide you through installing, configuring, and using vibe-check to test and automate your Claude Code agents.

Start by installing vibe-check in your project. This guide covers:

  • Package installation with npm/yarn/bun
  • Basic configuration
  • Environment setup for Claude Code
  • Verifying your installation

Time: 5 minutes Prerequisites: Node.js 18+, Claude Code CLI


The Quickstart guide introduces you to:

  • The dual API surface (vibeTest vs vibeWorkflow)
  • Agent definitions and execution
  • Auto-captured context (files, tools, git state)
  • Basic assertions and matchers

Time: 5 minutes Prerequisites: Installation complete


Follow Your First Test to learn:

  • Writing assertions with custom matchers
  • Inspecting file changes and tool calls
  • Using LLM judges for quality evaluation
  • Debugging failed tests with captured context

Time: 15 minutes Prerequisites: Quickstart complete


Learn to build workflows that:

  • Chain multiple agent stages together
  • Pass cumulative context between stages
  • Implement retry and loop patterns
  • Handle errors gracefully

Time: 20 minutes Prerequisites: Testing basics complete


Discover how to evaluate and benchmark agents:

  • Matrix testing across models and configurations
  • LLM-based quality evaluation with rubrics
  • Cost analysis and optimization
  • Comparing agent performance

Time: 20 minutes Prerequisites: Testing and automation basics complete


After completing the tutorials, explore:

  • How-To Guides - Task-oriented guides for specific use cases
  • API Reference - Complete API documentation for all functions and types
  • Explanation - Deep dives into architecture and design decisions