cloudbrain-cli 0.1.1

Infrastructure Cost Intelligence CLI — safe, read-only cloud cost optimization scanner for AWS and Databricks

cloudbrain

Infrastructure Cost Intelligence CLI — a safe, read-only scanner that identifies cloud infrastructure waste and generates optimization recommendations.

What it does

• Scans AWS and Databricks environments using read-only API access
• Identifies idle, underutilized, and misconfigured resources
• Estimates monthly savings potential with exact current vs recommended config
• Prioritizes recommendations by savings impact and risk level
• Scans: clusters, SQL warehouses, jobs, Unity Catalog, compute policies
• Outputs reports to terminal, JSON, or CSV

Install

pip install cloudbrain-cli

Usage

Scan

# Scan Databricks
export DATABRICKS_HOST="https://your-workspace.cloud.databricks.com"
export DATABRICKS_TOKEN="dapi..."
cloudbrain scan --provider databricks

# Scan AWS
export AWS_ACCESS_KEY_ID="your_key"
export AWS_SECRET_ACCESS_KEY="your_secret"
cloudbrain scan --provider aws

# Scan both providers
cloudbrain scan --provider aws --provider databricks

# Scan specific AWS regions
cloudbrain scan --provider aws --regions us-east-1,eu-west-1

Export

After running a scan, save the report without re-scanning:

# Save as JSON (to ~/Downloads)
cloudbrain export json

# Save as CSV (to ~/Downloads)
cloudbrain export csv

Commands

Command	Description
cloudbrain scan --provider <name>	Run a cost scan (results shown in terminal)
cloudbrain export json	Save last scan as JSON to ~/Downloads
cloudbrain export csv	Save last scan as CSV to ~/Downloads
cloudbrain init	Generate default config at ~/.cloudbrain/config.yaml
cloudbrain --version	Show version
cloudbrain --help	Show help

What it scans

Databricks

Resource	Checks
Clusters	Idle detection, auto-termination config, autoscaling config, oversized drivers
SQL Warehouses	Auto-stop config, sizing vs query rate, multi-cluster over-provisioning
Jobs	Failure rates, expensive jobs, long-running jobs, cost per run
Unity Catalog	Catalogs, schemas, Delta table optimization, external locations
Compute Policies	Missing policies, weak policies, clusters without policies

AWS

Resource	Checks
EC2	Idle instances (<5% CPU), underutilized instances, rightsizing suggestions
EBS	Orphaned volumes, old unattached snapshots
S3	Missing lifecycle policies, infrequent access candidates
ELB	Load balancers with no healthy targets
Elastic IPs	Unassociated IPs (charged hourly)
NAT Gateways	Low-traffic gateways

Security Model

• Read-only access only — never modifies your infrastructure
• No data exfiltration — all processing happens locally on your machine
• No database — nothing stored except a local cache of the last scan
• Least-privilege IAM — requires only Describe/List/Get permissions
• Credential safety — credentials never written to disk or logs

Configuration

# Generate default config
cloudbrain init

# Config location: ~/.cloudbrain/config.yaml

Configurable thresholds include:

• observation_window_days — how far back to look (default: 90)
• cpu_idle_percent — CPU threshold for idle classification (default: 5%)
• cpu_underutilized_percent — CPU threshold for rightsizing (default: 20%)
• cluster_idle_minutes — idle time before flagging (default: 30)
• warehouse_low_query_rate — queries/hour threshold (default: 5)

Example Output

💰 Estimated Monthly Savings: $16,704.87
📋 Total Findings: 148 (31 with cost impact)

━━━ Actionable Findings (31) ━━━

  #1 [MEDIUM RISK] OVERSIZED WAREHOUSE — $1,460.00/mo
      SQL Warehouse 'prod-serverless-analytics-sql-warehouse'
        ├─ Size: Small | Queries: 0.0/hr
        ├─ CURRENT CONFIG: cluster_size = Small ($4.00/hr)
        ├─ RECOMMENDED: cluster_size = X-Small ($2.00/hr)
        └─ ESTIMATED SAVINGS: $1460.00/mo

  #2 [LOW RISK] IDLE CLUSTER — $903.38/mo
      Cluster 'catalog-inventory-pipeline-compute'
        ├─ State: RUNNING (idle for 60 min)
        ├─ Node type: i4i.2xlarge × 2 workers
        ├─ Current cost: $1.24/hr ($903.38/mo if always on)
        ├─ Wasted so far this idle period: $1.23
        └─ ACTION: Terminate now or set auto-termination to 30 min

Requirements

• Python 3.11+
• AWS credentials (for AWS scanning) — configured via aws configure or environment variables
• Databricks PAT (for Databricks scanning) — generated from workspace settings

License

MIT