Autonomous AWS infrastructure monitoring — detects anomalies and self-heals before you even notice
CloudWatch Genius is an autonomous AI agent that transforms infrastructure monitoring from reactive firefighting into proactive, self-healing operations. Powered by Amazon Bedrock AgentCore and Claude 3 Sonnet, it detects anomalies in real time, executes safe remediation actions autonomously, and delivers executive-grade visibility — all without manual intervention.
- Advanced Anomaly Detection — Multi-algorithm approach (Z-score, seasonal decomposition, trend analysis) with sub-second detection and false-positive filtering
- Autonomous Remediation Engine — Safety-first design with multi-layered approval workflows, automatic rollbacks, and cooldown periods to prevent automation loops
- Enterprise Cost Optimization — AI-driven Reserved Instance recommendations, right-sizing analysis, and storage lifecycle policies targeting 25–40% cost savings
- Real-Time Executive Dashboard — CloudScape UI with live infrastructure health scores, anomaly timelines, cost trend projections, and automated weekly summaries
- Multi-Account & Multi-Region — Monitor 1,000+ AWS resources across unlimited accounts and regions from a single agent
- Audit-Ready Security — IAM least-privilege patterns, TLS 1.3 + AES-256 encryption, and full audit logging for every autonomous action
Amazon Bedrock AgentCore · Claude 3 Sonnet · Amazon CloudWatch · CloudScape UI · FastAPI · Plotly · Systems Manager · Auto Scaling · Cost Explorer API · SNS
# Clone and set up
git clone https://github.com/kyisaiah47/cloudwatch-genius
cd cloudwatch-genius
python3 -m venv venv && source venv/bin/activate
pip install fastapi uvicorn boto3 python-multipart
# Build frontend
cd frontend && npm install && npm run build && cd ..
# Launch dashboard
./venv/bin/python src/launcher.py --mode dashboard --port 8000Open http://localhost:8000 to access the dashboard.
Required environment variables:
AWS_REGION=us-east-1
AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
BEDROCK_MODEL_ID=anthropic.claude-3-sonnet-20240229-v1:0MIT