securevibes 0.1.1

AI-Native Platform to Secure Vibecoded Applications

🛡️ SecureVibes

AI-Native Security Scanner for Vibecoded Applications

SecureVibes uses Claude's multi-agent architecture to autonomously find security vulnerabilities in your codebase. Four specialized AI agents work together to deliver comprehensive, context-aware security analysis with concrete evidence.

---

🚀 Quick Start

# Install
pip install securevibes

# Configure API key
export CLAUDE_API_KEY="your-api-key-here"

# Scan your project
securevibes scan .

# View results
securevibes report

Get your Claude API key from: https://console.anthropic.com/

---

✨ What You Get

• ✅ Exact file paths and line numbers for every vulnerability
• ✅ CWE IDs for industry-standard tracking
• ✅ Concrete code snippets showing the vulnerable code
• ✅ Remediation recommendations with actionable fixes
• ✅ Exploitability analysis with realistic attack scenarios

---

🤖 Multi-Agent Architecture

SecureVibes orchestrates 4 specialized Claude agents:

1. Assessment Agent - Maps codebase architecture and technology stack
2. Threat Modeling Agent - Applies STRIDE methodology for realistic threats
3. Code Review Agent - Uses security thinking framework to find vulnerabilities
4. Report Generator - Compiles findings into actionable reports

Key Difference: Unlike traditional pattern-matching tools, SecureVibes agents understand your code's context, architecture, and business logic to find novel vulnerabilities that static analysis misses.

---

📊 Example Scan

$ securevibes scan .

🛡️ SecureVibes Security Scanner
📁 Scanning: /Users/xyz/repos/my-project
🤖 Model: sonnet
============================================================

✅ Phase 1/4: Architecture Assessment Complete
✅ Phase 2/4: Threat Modeling (STRIDE Analysis) Complete
✅ Phase 3/4: Code Review (Security Analysis) Complete
✅ Phase 4/4: Report Generation Complete

================================================================================
📊 Scan Results
================================================================================

  📁 Files scanned:   1953
  ⏱️  Scan time:       1053.66s
  💰 Total cost:      $2.27
  🐛 Issues found:    28
     🔴 Critical:     5
     🟠 High:         10
     🟡 Medium:       10
     🟢 Low:          3

---

🎯 Common Use Cases

# Export JSON for CI/CD pipeline
securevibes scan . --format json --output security-report.json

# Focus on critical/high severity
securevibes scan . --severity high

# Fast scan with cheaper model
securevibes scan . --model haiku

# Quiet mode for automation
securevibes scan . --quiet

---

⚙️ Configuration

Control agent models and analysis depth via environment variables:

# Required
export CLAUDE_API_KEY='your-api-key'

# Optional: Customize agent models (default: sonnet)
export SECUREVIBES_CODE_REVIEW_MODEL="opus"  # Max accuracy
export SECUREVIBES_THREAT_MODELING_MODEL="sonnet"

# Optional: Control analysis depth (default: 50)
export SECUREVIBES_MAX_TURNS=75  # Deeper analysis

Models: haiku (fast/cheap) | sonnet (balanced) | opus (thorough/expensive)

---

🐍 Python API

import asyncio
from securevibes import SecurityScanner

async def main():
    scanner = SecurityScanner(
        api_key="your-api-key",
        model="claude-3-5-sonnet-20241022"
    )
    
    result = await scanner.scan("/path/to/repo")
    print(f"Found {len(result.issues)} vulnerabilities")

asyncio.run(main())

---

🔒 Privacy & Security

What SecureVibes sends to Anthropic:

• Your source code files
• Relative file paths within the scanned repository

What SecureVibes does NOT send:

• Absolute paths or usernames
• Environment variables or secrets
• Git history or metadata

⚠️ Important: SecureVibes sends your code to Anthropic's Claude API for analysis. Review Anthropic's Privacy Policy before scanning proprietary code.

---

📚 Full Documentation

This is a quick reference for PyPI users. For comprehensive documentation, visit:

📖 Full Documentation on GitHub

Including:

• 🏗️ Architecture Deep Dive
• 🔧 Advanced Configuration
• 🧪 Example Scans & Results
• 🤝 Contributing Guide

---

👤 Author & Support

Built by @anshumanbh

• 🐛 Bug Reports: GitHub Issues
• 💡 Feature Requests: GitHub Discussions
• 🌟 Star the repo to follow development!

---

🙏 Acknowledgments

• Powered by Claude by Anthropic
• Built with Claude Agent SDK
• Inspired by traditional SAST tools, reimagined with AI

---

License: AGPL-3.0 | Requires: Python 3.10+