AntiVibe is an anti-vibecoding learning framework for Claude Code. It generates detailed, learning-focused explanations of AI-written code, helping you understand what AI writes - not just accept it.
# Clone this repository
git clone https://github.com/mohi-devhub/antivibe.git
# Copy to Claude Code global skills directory
cp -r antivibe ~/.claude/skills/antivibeUse the /antivibe command or describe what you want to learn:
/antivibe- Start a deep dive- "deep dive" - Analyze recently written code
- "learn from this code" - Generate learning guide
- "explain what AI wrote" - Explain specific files
- "understand what AI wrote" - Understand design decisions
Configure hooks to auto-trigger after task completion:
- Copy hooks configuration to your project:
cp hooks/hooks.json your-project/.claude/hooks.json- Or manually integrate into existing hooks:
{
"hooks": {
"SubagentStop": [...],
"Stop": [...]
}
}Generated deep-dives are saved to the deep-dive/ folder:
your-project/
├── deep-dive/
│ ├── auth-system-2026-01-15.md
│ ├── api-layer-2026-01-15.md
│ └── database-models-2026-01-15.md
Each file contains:
- Overview of what the code does
- Code walkthrough with explanations
- Concepts explained (design patterns, algorithms)
- Curated learning resources
- Related code in your project
- Next steps for deeper learning
Edit hooks/hooks.json to customize:
{
"hooks": {
"SubagentStop": [
{
"matcher": ".*",
"hooks": [...] // Remove to disable
}
],
"Stop": [
{
"matcher": ".*",
"hooks": [...] // Remove to disable
}
]
}
}Edit scripts/generate-deep-dive.sh to change output directory:
OUTPUT_DIR="your-custom-folder" # Default: "deep-dive"antivibe/
├── SKILL.md # Main skill definition
├── hooks/
│ └── hooks.json # Auto-trigger configuration
├── scripts/
│ ├── capture-phase.sh # Detect implementation phases
│ ├── analyze-code.sh # Parse code structure
│ ├── find-resources.sh # Find external resources
│ └── generate-deep-dive.sh # Generate markdown output
├── agents/
│ └── explainer.md # Subagent for detailed analysis
├── templates/
│ └── deep-dive.md # Output template
├── reference/
│ ├── language-patterns.md # Framework-specific patterns
│ └── resource-curation.md # Curated learning resources
└── docs/
└── setup.md # This file
- Trigger: You invoke
/antivibeor it triggers automatically - Identify: Find code written in the session (via git diff or explicit file list)
- Analyze: Use the explainer agent to deeply analyze the code
- Research: Find relevant external resources
- Generate: Create markdown output in
deep-dive/folder - Learn: Read the detailed explanation and follow linked resources
AntiVibe focuses on:
- Why over what - Explain design decisions
- Curated resources - Quality links, not random results
- Phase-aware - Group by implementation phase
- Learning paths - Suggest next steps
- Concept mapping - Connect to underlying CS principles
Input: "Explain the authentication system" Output:
# Deep Dive: Authentication System
## Overview
This auth system uses JWT tokens with refresh token rotation...
## Code Walkthrough
### auth/service.ts
- **Purpose**: Handles token generation and validation
- **Key Components**:
- `generateTokens()`: Creates access/refresh tokens
- `verifyToken()`: Validates JWT signatures
## Concepts Explained
### JWT (JSON Web Tokens)
- **What**: Stateless authentication tokens...
- **Why**: Server doesn't need to store sessions...
- **When**: APIs, SPAs, microservices...
## Learning Resources
- [JWT.io](https://jwt.io): Official documentation
- [Auth0 Guide](https://auth0.com/blog): Best practices- Ensure
SKILL.mdis in the correct location - Check that YAML frontmatter is valid
- Verify
namefield uses lowercase and hyphens only
- Verify
hooks.jsonis valid JSON - Ensure hooks are in project
.claude/folder - Check Claude Code version supports hooks
- Verify
deep-dive/folder exists - Check write permissions
- Ensure you're in a git repository
To extend AntiVibe:
- Add new patterns to
reference/language-patterns.md - Add resources to
reference/resource-curation.md - Customize the template in
templates/deep-dive.md
MIT