VibeKit VDK CLI

Stop Training Your AI Assistant Every Day

The world's first Vibe Development Kit - Train any AI coding assistant to understand your project like a senior developer who's been on your team for years.

VDK CLI revolutionizes AI-assisted development by generating intelligent rules that make AI suggestions 60% faster and perfectly matched to your codebase patterns. This tool helps developers create project-aware AI assistants and provides seamless integration across all major coding platforms.

Ecosystem Role (CLI is the Execution Core)

VDK CLI is the operational heart of the ecosystem:

1. ai-context-schema defines canonical blueprint contract
2. VDK-Blueprints provides curated inventory
3. VDK-CLI resolves, ranks, adapts, and deploys blueprints
4. VDK-Hub surfaces synchronized catalog/distribution UX
5. VDK-Wiki documents schema + CLI operational behavior

Hub and Wiki should mirror CLI retrieval/deployment semantics rather than redefine them.

Demo

Interactive Setup & Features

Quick Start Workflow

Features

Intelligent Project Analysis Automatically detects 20+ technologies, frameworks, and architectural patterns in your codebase

Universal AI Compatibility Works with 33+ IDEs and AI tools: Claude Code, Cursor, OpenCode, Cline, Roo Code, Goose, Junie, Google Antigravity, Kimi CLI, Mistral Vibe, Trae, VS Code variants, Windsurf, all JetBrains IDEs, Zed, GitHub Copilot, and more

Smart Rule Generation Creates tailored MDC files with project-specific patterns, conventions, and best practices

5-Minute Setup Zero configuration required - instant value with automatic project detection and rule deployment

Quick Start

Installation

# Install globally
npm install -g @vibe-dev-kit/cli

# Or use with pnpm
pnpm add -g @vibe-dev-kit/cli

# Or use the installer script
curl -fsSL https://raw.githubusercontent.com/vdkit/VibeKit-VDK-CLI/main/install.sh | sh

Basic Usage

# Initialize in your project
cd your-project
vdk init

# Interactive mode with guided setup
vdk init --interactive

# Check what VDK detected
vdk status

That's it! Your AI assistant now understands your project's patterns and conventions.

How It Works

1. Project Analysis: Scans your codebase to detect technologies, frameworks, and architectural patterns
2. Pattern Recognition: Identifies naming conventions, code structures, and project-specific practices
3. Rule Generation: Creates customized MDC files with intelligent guidelines for your tech stack
4. AI Integration: Deploys rules to your preferred AI assistant's configuration directory

Migration System

VDK CLI includes intelligent migration capabilities to convert your existing AI contexts from various platforms to VDK's universal format:

Supported Migration Sources

• Claude Code - CLAUDE.md files and .claude/ directories
• Cursor - .cursorrules and .cursor/ configurations
• GitHub Copilot - .github/copilot-instructions.md and related files
• Windsurf - .windsurf/ rules and configurations
• Generic AI - Various AI assistant rule formats

Migration Process

1. Detection: Automatically identifies existing AI contexts in your project
2. Analysis: Evaluates confidence levels and categorizes found contexts
3. Conversion: Transforms contexts to VDK blueprint format with project-specific adaptations
4. Organization: Creates structured migration output in vdk-migration/ folder
5. Integration: Optionally deploys converted rules to your configured AI assistants

Migration Commands

# Migrate existing AI contexts
vdk migrate

# Preview what would be migrated (no files created)
vdk migrate --dry-run

# Migrate without deploying to IDE integrations
vdk migrate --no-deploy

# Check for existing AI contexts before migrating
vdk status

The migration system leverages VDK's existing architecture (scanners, adapters, generators) to ensure converted rules maintain the same quality and project-awareness as newly generated ones.

Supported Technologies

Frontend Frameworks

Backend & Languages

IDEs & AI Assistants

Build Tools & Services

Core Commands

# Project initialization
vdk init                         # Initialize with auto-detection
vdk init --interactive          # Interactive setup with guided choices

# Migration
vdk migrate                      # Migrate existing AI contexts to VDK format
vdk migrate --dry-run           # Preview migration without creating files
vdk migrate --no-deploy         # Skip deployment to IDE integrations

# Analysis and status
vdk scan                         # Re-analyze project and update rules
vdk status                       # Check current VDK configuration
vdk validate                     # Validate generated rules

# Blueprint library operations
vdk sync                         # Sync blueprint sources
vdk browse                       # Browse available blueprints
vdk search --query nextjs        # Search canonical blueprint metadata
vdk deploy <blueprint-id>        # Deploy blueprint to active integrations
vdk repo-stats                   # Repository statistics and compliance summary

# AI Context Management
vdk generate --type main --name project-memory --platform claude-code   # Generate CLAUDE.md-compatible context
vdk generate --type rule --name cursor-rules --platform cursor          # Generate Cursor rule content
vdk import --input ./.claude --analyze                                  # Analyze existing Claude configuration
vdk import --input ./.cursor                                             # Import existing Cursor rules

# Cross-Platform Conversion
vdk convert --from claude --to cursor   # Convert Claude rules to Cursor
vdk convert --from cursor --to copilot  # Convert Cursor rules to Copilot

# Utilities
vdk --help                       # Show all available commands
vdk --version                    # Display VDK CLI version

Canonical Blueprint Retrieval (v3)

VDK CLI uses canonical AI Context Schema v3 metadata when searching and deploying from VDK-Blueprints.

• Canonical taxonomy filter: --kind
• Specificity filter: --specificity L0|L1|L2|L3|L4
• Curated default retrieval: blends L0-L3
• Provenance variants (L4) are excluded by default and only included with --include-l4

Curated library behavior

The default recommendation flow is intentionally curated to reduce noise:

• Start from generic + specific signal layers (L0-L3)
• Exclude provenance duplicates (L4) unless explicitly requested
• Preserve semantic adaptation outcome on conversion/deployment (lossless, lossy, unsupported)

# Search by canonical kind
vdk search --kind skill --platform claude-code

# Search by specificity layer
vdk search --query nextjs --specificity L2

# Include provenance variants explicitly
vdk search --query nextjs --include-l4

# Deploy a canonical blueprint by id/slug
vdk deploy <blueprint-id>

Supported IDEs & AI Tools

VDK CLI supports 33+ IDEs and AI assistants with full auto-detection and configuration:

• Code Editors: VS Code, VS Code Insiders, VSCodium, Cursor AI, Windsurf, Windsurf Next, Zed Editor
• JetBrains IDEs: IntelliJ IDEA, WebStorm, PyCharm, PHPStorm, RubyMine, CLion, DataGrip, GoLand, Rider, Android Studio
• AI Services: Claude Code, Claude Desktop, OpenAI Codex, OpenCode, Cline, Roo Code, Goose, Junie, Google Antigravity, Kimi CLI, Mistral Vibe, Trae, GitHub Copilot, OpenAI API Tools, Generic AI Platform
• Generic: Works with any AI coding assistant through universal .ai/ configuration

✨ ** Features**:

• Individual JetBrains IDE detection with project-specific configurations
• MCP (Model Context Protocol) integration across compatible platforms
• Confidence-based IDE detection with intelligent project analysis
• Automatic rule directory creation and management

📖 Complete IDE Support List - Exhaustive list with configuration details, MCP support status, and integration features.

Configuration

VDK CLI automatically generates configuration files for your detected AI assistants:

{
  "project": {
    "name": "my-nextjs-app",
    "type": "web-application",
    "frameworks": ["nextjs", "react", "typescript"],
    "libraries": ["tailwindcss", "shadcn-ui"]
  },
  "integrations": {
    "claude-code": {
      "enabled": true,
      "memoryFiles": ["CLAUDE.md", "CLAUDE-patterns.md"]
    },
    "cursor": {
      "enabled": true,
      "rulesDirectory": ".ai/rules/"
    }
  }
}

Examples

Basic Project Setup

# Navigate to your project
cd my-nextjs-app

# Initialize VDK
vdk init

# VDK detects: Next.js, React, TypeScript, Tailwind CSS
# Generates: .ai/rules/ with 15+ tailored rules
# Configures: Claude Code memory files

Interactive Setup

# Interactive mode for custom configuration
vdk init --interactive

# Choose your AI assistants:
# ✓ Claude Code
# ✓ Cursor
# ✗ Windsurf
# ✗ GitHub Copilot

# Generates optimized rules for selected platforms

Migration Example

# Project with existing AI contexts
cd my-existing-project

# Check what AI contexts are detected
vdk status
# Found: CLAUDE.md, .cursorrules, .github/copilot-instructions.md

# Preview migration without creating files
vdk migrate --dry-run
# Would migrate: 3 contexts with high confidence

# Perform migration
vdk migrate
# Migrated: 3 contexts → 15+ VDK blueprints in vdk-migration/
# Deployed: Rules to Claude Code, Cursor integrations

Advanced Usage

# Re-analyze after adding new dependencies
npm install prisma
vdk scan

# Check what VDK detected
vdk status

# Validate all generated rules
vdk validate

Blueprint contract quality commands

# Existing code quality (Biome + markdown)
pnpm run quality

# Blueprint frontmatter contract linting for .vdk/blueprints/rules
pnpm run lint:contracts

# Preview deterministic frontmatter normalization
pnpm run lint:contracts:dry

# Apply deterministic frontmatter normalization
pnpm run lint:contracts:fix

Real Impact

Teams using VDK CLI v2.0 report:

• 60% faster initial AI suggestions with enhanced technology detection
• 85% more relevant code completions with framework-specific rules
• 40% fewer back-and-forth clarifications through precise context
• 90% consistency in code patterns across team members
• 100% accuracy in package manager and build tool detection

Documentation

• Getting Started Guide - Complete setup instructions
• Supported IDEs & AI Tools - Complete compatibility list
• CLI Reference - All commands and options
• Rule Authoring Guide - Create custom rules
• Troubleshooting Guide - Common issues and solutions
• Scanner Documentation - Technical scanner details

Contributing

We welcome contributions! Please see our Contributing Guide for details.

• Report bugs
• Request features
• Improve documentation
• Submit pull requests

Roadmap

• [x] v2.0 - technology detection, improved AI integrations
• [x] v2.9 - AI context migration system with multi-platform support
• [ ] v2.2 - VDK Hub integration for cloud-based rule management
• [ ] v2.3 - Visual Studio Code extension for seamless IDE integration
• [ ] v3.0 - Team collaboration features and shared rule repositories

Requirements

• Node.js: >= 22.0.0
• npm: >= 8.0.0 or pnpm >= 7.0.0
• Operating System: Windows, macOS, Linux

License

MIT License - see the LICENSE file for details.

Project Evolution & Attribution

VDK CLI is the evolution of a rich history of AI-assisted development tooling:

• Original DevRules by Seth Rose - Established the foundational concept of structured AI assistant rules
• AIRules - Major enhancement adding 51+ specialized tasks, memory management, and multi-platform support
• VibeKit VDK - The current evolution, a comprehensive toolkit for making AI assistants project-aware

For detailed history and contributions, see Project History & Attribution.

Support

• GitHub: Issues & Feature Requests
• Documentation: Complete Docs
• Community: VDK Hub
• Email: For enterprise support inquiries

GitHub • Documentation • VDK Hub • NPM