Cloud cost optimization engine for AWS and Azure.

Scans your infrastructure, finds wasted resources, and reports exact monthly and annual savings per resource.

What it detects

Category	AWS	Azure
Compute	Idle EC2 instances, oversized instance types	Idle VMs
Storage	Unattached EBS volumes, S3 lifecycle gaps, old versions	Unattached managed disks
Network	Idle load balancers with no healthy targets	Unassociated public IPs
Database	Idle RDS instances, unnecessary Multi-AZ deployments	—

Architecture

cloudcostguard scan --provider aws
         |
         v
    [Engine]               Concurrent scanner orchestration
     /      \
[AWS]      [Azure]         Provider-specific scanners
  |          |
  EC2  EBS   VM  Disk      Resource scanners (per type)
  S3   ELB   Network
  RDS
         |
         v
    [Pricing]              Per-resource cost lookup + region multipliers
         |
         v
  [Report]                 Table / JSON / CSV output with $ savings

Quick start

# Install
go install github.com/Aliipou/cloudcostguard@latest

# Or build from source
git clone https://github.com/Aliipou/cloudcostguard.git
cd cloudcostguard && make build

# Configure
cp cloudcostguard.example.yaml ~/.cloudcostguard.yaml
# Edit with your subscription ID / AWS profile

# Scan
cloudcostguard scan --provider aws
cloudcostguard scan --provider azure --type compute
cloudcostguard scan --provider aws --min-savings 50 --output json

# Web dashboard + Prometheus metrics
cloudcostguard serve --port 8080
# Open http://localhost:8080

Output formats

Table (default):

SEVERITY   PROVIDER   CATEGORY   TITLE                                      MONTHLY       ANNUAL   EFFORT
--------------------------------------------------------------------------------------------------------------
CRITICAL   aws        compute    Idle EC2 instance: api-server-prod           $560.64    $6727.68   low
HIGH       aws        storage    Unattached EBS volume: old-data (500GB)      $50.00      $600.00   low
MEDIUM     azure      compute    Idle Azure VM: staging-web                   $70.08      $840.96   low
--------------------------------------------------------------------------------------------------------------
TOTAL (3 findings)                                                            $680.72    $8168.64

JSON — for automation and pipelines:

cloudcostguard scan --provider aws --output json | jq '.summary.total_annual_savings'

CSV — for spreadsheets:

cloudcostguard scan --provider aws --output csv > report.csv

Configuration

# ~/.cloudcostguard.yaml
provider: aws

aws:
  profile: default
  regions: [us-east-1, us-west-2, eu-west-1]

rules:
  idle_cpu_threshold: 5.0       # % CPU below = idle
  idle_days: 14                 # observation window in days
  unattached_disk_days: 7       # days before flagging unattached disks
  oversized_cpu_threshold: 20.0 # % CPU below = oversized

See cloudcostguard.example.yaml for all options.

Monitoring

Start the full stack (app + Prometheus + Grafana):

docker compose up -d

Service	URL
Dashboard	http://localhost:8080
Prometheus	http://localhost:9090
Grafana	http://localhost:3000

Grafana is pre-provisioned with a CloudCostGuard dashboard showing findings by severity, savings by category, and scan duration histograms.

Required permissions

AWS — attach this read-only policy:

{
  "Effect": "Allow",
  "Action": [
    "ec2:DescribeInstances", "ec2:DescribeVolumes",
    "s3:ListAllMyBuckets", "s3:GetBucketLocation", "s3:GetLifecycleConfiguration",
    "elasticloadbalancing:DescribeLoadBalancers", "elasticloadbalancing:DescribeTargetHealth",
    "rds:DescribeDBInstances",
    "cloudwatch:GetMetricStatistics"
  ],
  "Resource": "*"
}

Azure — Reader role on the subscription is sufficient.

Development

make test    # run tests with coverage
make lint    # golangci-lint
make build   # build binary
make docker  # build Docker image

Project structure

cmd/                    CLI entry points (scan, serve, version)
internal/
  api/                  HTTP server + web dashboard handler
  config/               YAML config loading with defaults
  engine/               Concurrent scanner orchestration
  metrics/              Prometheus-compatible metrics (stdlib only)
  model/                Finding, Severity, Category types
  pricing/              Cloud pricing tables + region multipliers
  report/               Table, JSON, CSV formatters
  scanner/
    aws/                EC2, EBS, S3, ELB, RDS scanners
    azure/              VM, Disk, Network scanners
web/                    Embedded single-file dashboard
monitoring/             Prometheus + Grafana configs

License

MIT