CVE Scanning and Remediation System

Automated CVE scanning with AI-powered parallel remediation agents.

Features

• Trivy CVE Scanning: Scans repositories for known vulnerabilities
• Parallel AI Agents: Multiple agents fix vulnerabilities simultaneously
• Real-time Status: Shows actual agent actions ("Editing file package.json", "Running: npm install")
• Auto Pull Requests: Creates PRs for each vulnerability fix
• GitHub Integration: Works with public and private repositories

Quick Start

1. Install dependencies:

pip install -r requirements.txt

2. Set API key:

export LLM_API_KEY="your-anthropic-api-key"
export GITHUB_TOKEN="your-github-token"  # optional

3. Run scan and fix:

python cve_scan.py https://github.com/owner/repo.git

Configuration

• LLM_API_KEY: Anthropic API key (required)
• GITHUB_TOKEN: GitHub token for private repos/PRs (optional)
• MAX_PARALLEL_AGENTS: Parallel agents (default: 3)
• STATUS_UPDATE_INTERVAL: Update frequency in seconds (default: 5)

How It Works

1. Scan: Trivy scans repository → vulnerabilities.json
2. Fix: AI agents work in parallel to fix each vulnerability
3. PR: Each agent creates a pull request with the fix

Example Output

🔍 PHASE 1: CVE SCANNING
✅ Found 3 vulnerabilities in 8.2s

🔧 PHASE 2: REMEDIATION  
🤖 Agent-1 (CVE-2023-1234): Editing file package.json
🤖 Agent-2 (CVE-2023-5678): Running: npm audit fix
🤖 Agent-3 (CVE-2023-9012): Creating security patch

✅ All agents completed: 3 PRs created

Requirements

• Python 3.8+
• Docker (for Trivy)
• Anthropic API key