finops-ai 0.1.1

AI-powered multi-cloud cost optimization for humans

FinOps AI 🤖☁️

Enterprise-Grade Multi-Cloud Cost Optimization Platform

Features • Quickstart • Providers • AI Engine • Configuration • Contributing

---

Stop paying for cloud resources nobody uses. FinOps AI scans your multi-cloud infrastructure, detects orphaned resources, estimates waste, and provides AI-powered recommendations — saving enterprises an average of 30% on cloud spend.

🚀 Features

Feature	Description
Multi-Cloud Scanning	Azure, AWS, and GCP — one unified interface
14+ Resource Types	Snapshots, disks, VMs, IPs, NICs, LBs, RDS, App Services, and more
AI/ML Engine	Isolation Forest anomaly detection + Prophet cost forecasting
Smart Recommendations	Prioritized, actionable savings with effort/risk ratings
Policy Engine	YAML-driven governance with approval workflows
Dependency Graph	NetworkX-powered safe-deletion analysis
Rich Reports	HTML (dark theme), JSON, CSV, and Slack notifications
Beautiful CLI	Click + Rich with color-coded output and progress bars

⚡ Quickstart

Install

# Core package
pip install finops-ai

# With cloud providers
pip install finops-ai[azure]     # Azure only
pip install finops-ai[aws]       # AWS only
pip install finops-ai[gcp]       # GCP only
pip install finops-ai[all]       # All providers + ML

# Development
pip install finops-ai[dev]

First Scan

# Generate a config file
finops-ai init

# Scan Azure (uses CLI auth by default)
finops-ai scan --provider azure

# Scan all providers
finops-ai scan --provider all

# Scan with JSON report
finops-ai scan --provider azure --output json --report my_report.json

# Check version and installed providers
finops-ai version

Python API

from finops_ai.core.auth_manager import AuthManager
from finops_ai.providers.azure.snapshot_manager import AzureSnapshotManager

# Authenticate
credential = AuthManager.get_azure_credential()

# Scan for orphaned snapshots
manager = AzureSnapshotManager(credential)
result = manager.scan()

print(f"Found {result.total_resources} orphaned snapshots")
print(f"Estimated savings: ${result.total_estimated_cost:.2f}/month")

for resource in result.resources:
    print(f"  {resource.name}: ${resource.estimated_monthly_cost:.2f}/mo ({resource.age_days} days old)")

☁️ Providers

Azure (7 Resource Types)

• Snapshots — Orphaned managed snapshots (source disk deleted)
• Disks — Unattached managed disks (with SKU-aware pricing)
• Network — Unused public IPs, detached NICs, idle load balancers
• VMs — Zombie VMs (stopped/deallocated > N days)
• Storage — Idle blob containers (>90 days no modification)
• App Service Plans — Plans with zero web apps
• Resource Groups — Empty resource groups

AWS (4 Resource Types)

• EC2 — Stopped instances (zombie VMs) + unused security groups
• EBS — Unattached volumes + orphaned snapshots (source volume deleted)
• Network — Unassociated Elastic IPs + idle ALBs/NLBs
• RDS — Orphaned manual snapshots (source DB deleted)

GCP (3 Resource Types)

• Compute — Orphaned snapshots + unattached disks + stopped VMs
• Network — Reserved (unused) static IPs (regional + global)
• Storage — Empty Cloud Storage buckets

🧠 AI Engine

from finops_ai.ml.anomaly_detector import AnomalyDetector
from finops_ai.ml.cost_forecaster import CostForecaster
from finops_ai.ml.recommender import SmartRecommender

# Anomaly detection (Isolation Forest)
detector = AnomalyDetector(contamination=0.1)
anomalies = detector.detect(resources)

# Cost forecasting (Linear/Prophet)
forecaster = CostForecaster()
forecast = forecaster.forecast_from_snapshots(historical_data, forecast_days=30)

# Smart recommendations
recommender = SmartRecommender()
report = recommender.analyze(resources)
print(f"Total savings: ${report.total_annual_savings:,.2f}/year")

⚙️ Configuration

Generate a config file with finops-ai init, then customize:

dry_run: true
log_level: INFO

providers:
  azure:
    enabled: true
    auth:
      method: cli  # cli | managed_identity | service_principal
    subscription_id: ""

  aws:
    enabled: true
    auth:
      method: profile
    regions: [us-east-1, us-west-2]

  gcp:
    enabled: true
    auth:
      method: adc
    project_id: my-project

policies:
  policies_dir: ./policies
  enforce: false

ml:
  enabled: true
  anomaly_detection: true
  cost_forecasting: true

🏗️ Architecture

src/finops_ai/
├── __init__.py              # Package root
├── cli.py                   # Click + Rich CLI
├── config.py                # Pydantic Settings configuration
├── core/
│   ├── auth_manager.py      # Multi-cloud authentication
│   ├── base_manager.py      # Abstract base + data models
│   ├── graph_analyzer.py    # Dependency graph (NetworkX)
│   └── policy_engine.py     # YAML governance policies
├── providers/
│   ├── azure/               # 7 Azure resource managers
│   ├── aws/                 # 4 AWS resource managers
│   └── gcp/                 # 3 GCP resource managers
├── ml/
│   ├── anomaly_detector.py  # Isolation Forest
│   ├── cost_forecaster.py   # Prophet / linear regression
│   └── recommender.py       # Smart recommendations
├── reporters/
│   ├── json_reporter.py     # JSON export
│   ├── csv_reporter.py      # CSV export
│   ├── html_reporter.py     # Standalone dark-theme HTML
│   └── slack_reporter.py    # Slack webhook (Block Kit)
└── utils/
    ├── logger.py             # Rich logging + JSON audit
    └── cost_calculator.py    # Cross-cloud pricing

🧪 Testing

# Run tests
pytest tests/ -v

# With coverage
pytest tests/ --cov=finops_ai --cov-report=html

📋 Migrating from TerraSnap-Govern

The original TerraSnap-Govern code has been preserved in legacy/. To migrate:

# Old way
python scripts/azure_snapshot_cleanup.py --subscription-id $SUB_ID

# New way
finops-ai scan --provider azure --resource-type snapshot -s $SUB_ID

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/my-feature
3. Install dev dependencies: pip install -e ".[dev]"
4. Make changes and add tests
5. Run linting: ruff check src/
6. Run tests: pytest tests/ -v
7. Submit a pull request

📄 License

MIT License — see LICENSE for details.

---

Built with ❤️ for FinOps practitioners everywhere