Multi-Agent AI Code Analysis System powered by Google ADK and Cloud Run

An enterprise-grade AI-powered code analysis platform that uses multi-agent architecture to provide comprehensive code intelligence. Built for the Google Cloud Run Hackathon 2025 in the AI Agents Category.

• 🤖 Multi-Agent AI System - 4 specialized agents working sequentially
• 🔒 Security Analysis - SQL injection, XSS, and vulnerability detection
• ⚡ Performance Optimization - Complexity analysis and bottleneck detection
• 🔧 Code Refactoring - Best practices and code quality improvements
• 📊 Real-time Dashboard - System metrics and analysis tracking
• ☁️ Cloud-Native - Fully serverless on Google Cloud Run

┌─────────────────────────────────────────────────────────┐
│                   User Code Input                        │
└─────────────────┬───────────────────────────────────────┘
                  │
                  ▼
┌─────────────────────────────────────────────────────────┐
│              Multi-Agent Coordinator                     │
│           (Sequential Execution Pattern)                 │
└─────────────────┬───────────────────────────────────────┘
                  │
        ┌─────────┴─────────┬─────────┬─────────┐
        ▼                   ▼         ▼         ▼
   ┌────────┐         ┌──────────┐ ┌────────┐ ┌────────┐
   │ Parser │ ──────> │ Security │ │  Perf  │ │Refactor│
   │ Agent  │         │  Agent   │ │ Agent  │ │ Agent  │
   └────────┘         └──────────┘ └────────┘ └────────┘
        │                   │         │         │
        └─────────┬─────────┴─────────┴─────────┘
                  ▼
┌─────────────────────────────────────────────────────────┐
│            Combined Analysis Results                     │
└─────────────────────────────────────────────────────────┘

Backend:

• FastAPI (Python 3.11)
• Google Generative AI (Gemini 2.0 Flash)
• Google Cloud Run (8 vCPU, 8GB RAM)
• Google Firestore (NoSQL Database)
• Google Cloud Storage (File Storage)
• Google Secret Manager (API Key Management)

Frontend:

• React 18.2 + TypeScript
• Ant Design 5.x
• Vite 5.0
• Nginx (Production Server)

• Google Cloud Project with billing enabled
• Google Cloud SDK (gcloud) installed
• Gemini API Key
• Node.js 18+ and Python 3.11+

Backend:

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

export GEMINI_API_KEY="your-api-key"
export PROJECT_ID="your-project-id"
export REGION="us-central1"

uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

Frontend:

cd frontend
npm install
npm run dev

Backend:

cd backend
gcloud run deploy code-intelligence-api \
  --source . \
  --region us-central1 \
  --platform managed \
  --allow-unauthenticated \
  --memory 8Gi \
  --cpu 8 \
  --timeout 300 \
  --min-instances 2 \
  --max-instances 20 \
  --set-env-vars PROJECT_ID=your-project-id,REGION=us-central1 \
  --update-secrets GEMINI_API_KEY=GEMINI_API_KEY:latest

Frontend:

cd frontend
gcloud run deploy code-intelligence-ui \
  --source . \
  --region us-central1 \
  --platform managed \
  --allow-unauthenticated \
  --memory 512Mi \
  --cpu 1 \
  --timeout 60

Endpoint: POST /api/adk/analyze

Request:

{
  "code": "def get_user(id):\n    query = f\"SELECT * FROM users WHERE id = {id}\"\n    return db.execute(query)",
  "agent_type": "multi"
}

Response:

{
  "analysis_id": "uuid",
  "status": "completed",
  "results": {
    "parser": "Code structure analysis...",
    "security": "SQL injection vulnerability detected...",
    "performance": "Query optimization needed...",
    "refactoring": "Use parameterized queries..."
  },
  "timestamp": "2025-11-10T12:00:00Z"
}

Endpoint: GET /api/monitoring/health

Response:

{
  "status": "healthy",
  "queue": {
    "total_queued": 0,
    "executing": 0,
    "available_slots": 10
  },
  "system": {
    "cpu": 0.0,
    "memory": 1.4,
    "load_score": 0.42
  }
}

• Code structure analysis
• Syntax validation
• Complexity metrics (Cyclomatic, Halstead)
• Function/class detection

• SQL injection detection
• XSS vulnerability scanning
• Insecure deserialization
• Hardcoded credentials detection

• Time complexity analysis (O(n), O(n²), etc.)
• Memory usage optimization
• Database query optimization
• Caching recommendations

• Code quality improvements
• Best practices enforcement
• Design pattern suggestions
• Clean code recommendations

Real-time monitoring dashboard showing:

• System metrics (CPU, Memory, Load)
• Queue status (Queued, Executing, Completed)
• Performance statistics (P50, P95, P99 latency)
• Agent execution metrics

This project is built for the AI Agents Category of the Google Cloud Run Hackathon.

Why This Project Wins:

1. Multi-Agent Architecture - Demonstrates advanced AI agent coordination
2. Cloud-Native Design - Fully serverless, auto-scaling, cost-efficient
3. Production-Ready - Enterprise-grade monitoring, error handling, testing
4. Real-World Impact - Solves actual developer pain points in code review
5. Google Cloud Integration - Uses Cloud Run, Firestore, Storage, Secret Manager

MIT License

This is a hackathon project. Contributions are welcome after the competition ends.

Built with ❤️ for Google Cloud Run Hackathon 2025

#CloudRunHackathon #GoogleADK #AIAgents