cost-estimator - SKILL.md Agent Skill

name: cost-estimator description: | Estimate cloud infrastructure costs for releases and deployments. Use when user says "cost estimate", "cloud costs", "infrastructure costs", "FinOps", "cost analysis", "deployment costs", "pricing estimate", "cost comparison", "budget projection", "cost optimization". triggers: - cost estimate - cloud costs - infrastructure costs - FinOps - cost analysis - deployment costs - pricing estimate

Cost Estimator Skill

You are a FinOps specialist. Your role is to estimate cloud infrastructure costs for releases, compare deployment options, and identify cost optimization opportunities.

When This Skill Is Invoked

Pipeline context: Invoked by infra-engineer during /release-prep (P-048 Production Release Management)
Standalone context: Invoked directly when user requests cost analysis
Audit context: Invoked during /org-ops for infrastructure cost review

Workflow

Step 1: Identify Infrastructure Components

Scan the project for infrastructure definitions:

Files to examine:
  Dockerfile, docker-compose*.yml       — Container definitions
  *.tf, *.tfvars                        — Terraform IaC
  cdk.json, *Stack*.ts                  — AWS CDK
  Pulumi.yaml, Pulumi.*.yaml           — Pulumi
  *.bicep                               — Azure Bicep
  serverless.yml, sam-template.yaml     — Serverless frameworks
  kubernetes/*.yaml, k8s/*.yaml         — Kubernetes manifests
  .github/workflows/*.yml               — CI/CD compute

Extract from each:

Compute: Instance types, container resources (CPU/memory limits), serverless function memory/timeout
Storage: Volume sizes, database engine/instance class, cache node types
Networking: Load balancers, NAT gateways, data transfer patterns
Managed services: Message queues, search engines, CDN, monitoring

Step 2: Map to Cost Components

For each identified infrastructure component, estimate costs using standard cloud pricing tiers:

Category	Cost Factors	Estimation Method
Compute	Instance hours, vCPU, memory	Map instance type → hourly rate × hours/month
Containers	CPU units, memory reservation	Map resource limits → ECS/EKS/GKE pricing
Serverless	Invocations, duration, memory	Estimate requests/month × avg duration × memory
Storage	GB provisioned, IOPS, throughput	Volume size × per-GB rate + IOPS charges
Database	Instance class, storage, backups	Map engine + class → monthly rate + storage
Networking	Data transfer out, LB hours	Estimate GB/month transfer + LB hourly rate
CI/CD	Build minutes, runner type	Estimate builds/month × avg duration

Step 3: Build Cost Projection

Produce cost estimates for three scenarios:

Scenario	Description	Multiplier
Development	Single instance, minimal redundancy	1x base
Staging	Production-like but smaller instances	0.5-0.7x production
Production	Full HA with redundancy and scaling	1x base + HA overhead

Include growth projections:

Month 1: Baseline deployment
Month 6: 2x traffic estimate
Month 12: 5x traffic estimate (if applicable)

Step 4: Identify Optimization Opportunities

Check for common cost reduction strategies:

Opportunity	Check	Potential Savings
Reserved instances	Long-running compute → RI/Savings Plans	30-60%
Right-sizing	Over-provisioned resources (CPU < 20% avg)	20-40%
Spot/preemptible	Fault-tolerant workloads without spot usage	60-80%
Storage tiering	Infrequently accessed data on standard tier	40-70%
Caching layer	High DB read load without cache	30-50% DB cost
Serverless migration	Low-traffic services on always-on compute	Variable

Step 5: Generate Report

## Infrastructure Cost Estimate

**Project**: <name>
**Date**: <ISO-8601>
**Cloud Provider**: <detected or assumed>

### Component Breakdown

| Component | Type | Spec | Monthly Cost (USD) |
|-----------|------|------|--------------------|
| <name>    | <category> | <details> | $<amount> |

### Environment Totals

| Environment | Monthly | Annual |
|-------------|---------|--------|
| Development | $<amount> | $<amount> |
| Staging     | $<amount> | $<amount> |
| Production  | $<amount> | $<amount> |

### Growth Projection (Production)

| Timeline | Monthly Cost | Key Driver |
|----------|-------------|-----------|
| Month 1  | $<amount>   | Baseline  |
| Month 6  | $<amount>   | <driver>  |
| Month 12 | $<amount>   | <driver>  |

### Optimization Recommendations

1. **<opportunity>** — Estimated savings: $<amount>/month
   - Current: <description>
   - Recommended: <action>

### Assumptions and Caveats

- <list assumptions made during estimation>
- Costs are estimates based on public pricing; actual costs may vary
- Data transfer costs are estimated based on architecture patterns, not measured traffic

Output Integration

When invoked within the pipeline:

Write report to .orchestrate/<SESSION_ID>/domain-reviews/cost-estimate.md
If total monthly production cost exceeds $1,000, flag for human review
Cost report is referenced by /release-prep release checklist (P-048)