Analyze and Optimize Cloud Spending

The Cost Optimization Analyzer is an autonomous agent expert in cloud cost optimization and FinOps practices, specializing in analyzing infrastructure spending patterns, identifying cost anomalies, and delivering actionable recommendations to reduce cloud spend across AWS, Azure, and Google Cloud Platform. Its resource utilization assessment analyzes CPU, memory, storage, and network usage metrics, identifies over-provisioned resources with persistently low utilization (under 30%), detects idle resources at 0% utilization for extended periods, and calculates unit-cost metrics for comparison. Its right-sizing methodology uses 95th-percentile utilization for CPU and memory recommendations, applies 10-20% safety buffers for production workloads, and accounts for seasonal patterns and growth trends. It covers AWS-specific strategies (Reserved Instance and Savings Plan analysis calculating current versus RI-covered annual cost, Spot Instance optimization for fault-tolerant workloads with typical 70-90% savings versus On-Demand), storage cost optimization (S3 lifecycle policies transitioning objects through Standard-IA, Glacier, and Deep Archive storage classes, database storage type selection like gp2 versus gp3 versus io1), and cross-cloud cost comparison by mapping equivalent services (EC2/Virtual Machines/Compute Engine, S3/Blob Storage/Cloud Storage, RDS/SQL Database/Cloud SQL) across providers.

Use this agent when you need statistically grounded cost anomaly detection and right-sizing recommendations rather than a one-time manual spend review - it implements rolling averages, standard-deviation thresholds, seasonal decomposition, and z-score outlier detection with a configurable threshold (severity classified as high above a z-score of 3, medium otherwise) rather than eyeballing a cost dashboard. It produces monthly executive summaries with total cloud cost, month-over-month change, identified and realized savings, and a ranked action plan. It is not meant to push optimizations straight to production - its own risk-management guidance is to test optimizations in non-production environments first, maintain rollback procedures, and document performance-impact assessments before acting on any recommendation.

### Example S3 Lifecycle Policy
LifecycleConfiguration:
  Rules:
    - Id: OptimizeStorageCosts
      Status: Enabled
      Filter:
        Prefix: logs/
      Transitions:
        - Days: 30
          StorageClass: STANDARD_IA
        - Days: 90
          StorageClass: GLACIER
        - Days: 365
          StorageClass: DEEP_ARCHIVE
      Expiration:
        Days: 2555  # 7 years retention

Given cloud usage and pricing data, the agent outputs RI/Savings Plan savings calculations (current annual cost, RI annual cost, savings percentage), storage lifecycle policies like the one above, cross-cloud service cost comparisons, a cost anomaly report flagging dates, actual versus expected cost, deviation percentage, and severity, a KPI dashboard breaking down costs by service/team/environment with month-over-month and year-over-year growth rates, and a monthly executive summary with a top-ranked recommendation list (each tagged with a dollar-per-month savings estimate) and a checklist-style action plan.

FinOps practitioners and cloud infrastructure teams managing multi-cloud or single-cloud spend who need quantified, statistically-backed recommendations rather than ad hoc cost cutting. It suits organizations that want automation baked into the process - Infrastructure as Code for consistent provisioning, native cost tools like AWS Cost Explorer or Azure Cost Management, automated right-sizing recommendations with approval workflows, cost-allocation tagging, weekly cost-review meetings, and proactive budget alerts - and that treat cost changes with the same rollback and performance-impact discipline as any other infrastructure change.