cloudthinker-ai Cloud Cost Optimization Report Skill
Generate a structured cost optimization report for {{ cloud_provider | uppercase }} covering {{ time_period | "last 30 days" }}.
Workflow
Step 1 — Establish Cost Baseline
Gather current spending data:
- Total monthly spend — current month vs prior 3 months
- Cost breakdown by service — top 10 services by spend
- Cost by team/account/project — identify top-spending business units
- Month-over-month trend — is spend growing, stable, or declining?
- Budget vs actuals — are budgets being exceeded?
Expected output from this step:
COST BASELINE — {{ cloud_provider | uppercase }}
Period: {{ time_period }}
Total Spend: $X,XXX/month
MoM Change: +/-X%
Top Services:
1. [Service]: $X,XXX (XX%)
2. [Service]: $X,XXX (XX%)
...
Step 2 — Idle & Unused Resource Detection
For each service category, identify resources with zero or near-zero utilization:
Compute:
- Stopped/terminated EC2/GCE/VMs still incurring storage costs
- EC2 instances with CPU < 5% over 14 days
- Auto Scaling Groups at minimum capacity 24/7 (no scaling events)
Storage:
- Unattached EBS volumes / Persistent Disks
- Snapshots older than 90 days without recent access
- S3 buckets with no GET requests in 30 days (excluding logging buckets)
- Empty storage containers / buckets
Networking:
- Elastic IPs not associated with running instances
- Idle NAT Gateways (< 100MB/day transfer)
- Unused Load Balancers (0 healthy targets or 0 requests)
- Unattached Elastic Network Interfaces
Database:
- RDS/Cloud SQL instances with CPU < 2% and zero connections for 7 days
- Read replicas with zero replica lag and zero queries
Other:
- Unused Elastic Container Registries with images > 1 year old
- CloudFormation stacks in ROLLBACK state
- Forgotten development environments running in production accounts
Threshold for waste reporting: ${{ cost_threshold | "50" }}/month minimum
Step 3 — Rightsizing Analysis
Identify over-provisioned resources:
Compute Rightsizing:
For each compute instance with metrics:
- Average CPU over 14 days < 20% → downsize
- Average memory over 14 days < 30% → downsize
- Max CPU burst < 50% → no need for burstable exclusion
- Calculate estimated savings from downsizing
Database Rightsizing:
For each database instance:
- Average CPU < 15% → eligible for smaller instance class
- FreeableMemory average > 75% of allocated → memory over-provisioned
- IOPS average < 30% of provisioned → IOPS over-provisioned (if io1/io2)
Step 4 — Commitment / Reserved Capacity Analysis
Evaluate savings from longer-term commitments:
On-Demand vs Reserved Instance coverage
- What % of compute is on-demand vs reserved?
- Which on-demand instances are running 24/7 (RI candidates)?
- Reserved Instance utilization — are purchased RIs being used?
Savings Plan analysis
- What is current Savings Plan coverage?
- Recommended Savings Plan commitment amount
Commitment recommendation:
- Stable workloads running > 80% of time → 1-year reserved/committed
- Very stable workloads → 3-year for maximum savings
- Variable workloads → on-demand or spot-eligible
Step 5 — Spot / Preemptible Opportunities
Identify workloads that can run on spot/preemptible instances:
- Batch processing jobs
- CI/CD build agents
- Dev/test environments
- Stateless application tier (with proper interruption handling)
- ML training workloads
Estimated savings: 60-80% vs on-demand pricing
Step 6 — Data Transfer & Network Costs
Analyze network spend:
- Cross-region data transfer costs
- NAT Gateway usage vs VPC endpoints (S3, DynamoDB)
- CloudFront vs direct S3 serving for public assets
- VPC endpoint usage opportunities (eliminate NAT for AWS service traffic)
Step 7 — Anomaly Detection
Identify unusual cost spikes:
- Services with >20% MoM increase without corresponding business growth
- New services with unexpected costs (developer experiments left running)
- Data transfer spikes (potential data exfiltration or misconfigured logging)
- Unexpected region usage
Step 8 — Prioritized Recommendations
Rank all findings by impact and implementation effort:
SAVINGS RECOMMENDATIONS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
QUICK WINS (implement this week) — Low risk, immediate savings
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
1. [Resource]: [Description]
Monthly savings: $X,XXX | Risk: Low | Effort: Hours
Action: [specific command or step]
2. [Resource]: [Description]
Monthly savings: $XXX | Risk: Low | Effort: Minutes
Action: [specific command or step]
MEDIUM TERM (implement this month) — Some coordination needed
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
3. Rightsizing [service]: [X] instances downsize from [type] to [type]
Monthly savings: $X,XXX | Risk: Medium | Effort: Days
Action: [specific steps with validation approach]
4. Reserved Instances: Purchase [X] RIs for [service]
Monthly savings: $X,XXX | Risk: Low | Effort: Hours (approval needed)
Action: [purchase via console/API]
STRATEGIC (this quarter) — Requires architectural changes
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
5. [Architectural change]: [description]
Monthly savings: $X,XXX | Risk: High | Effort: Weeks
Action: [high-level steps and considerations]
Step 9 — Executive Summary
Produce a one-page summary:
COST OPTIMIZATION REPORT — {{ cloud_provider | uppercase }}
Generated: [date]
Period: {{ time_period }}
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CURRENT STATE
• Monthly Spend: $X,XXX
• MoM Trend: [+/-X%]
• Top Cost Driver: [service]
SAVINGS IDENTIFIED
• Quick Wins: $X,XXX/month (implement in <1 week)
• Medium Term: $X,XXX/month (implement in <1 month)
• Strategic: $X,XXX/month (implement in <1 quarter)
• Total Potential: $XX,XXX/month (XX% reduction)
TOP 3 RECOMMENDATIONS
1. [action]: $X,XXX/month
2. [action]: $X,XXX/month
3. [action]: $X,XXX/month
NEXT STEPS
• [Owner]: [Action] by [Date]
• [Owner]: [Action] by [Date]
Counter-Rationalizations
| Shortcut | Counter | Why |
|---|---|---|
| "We can skip some steps for this case" | Adapt the workflow steps, don't skip them | Skipped steps are where incidents and oversights originate |
| "The user seems to already know what to do" | Complete all workflow phases with the user | The workflow catches blind spots that experience alone misses |
| "This is a minor case, full process is overkill" | Scale the process down, don't turn it off | Minor cases become major when unstructured; the process scales, not disappears |
| "I'll fill in the details later" | Complete each section before moving on | Deferred details are forgotten; real-time capture is more accurate |
| "The template output isn't necessary" | Always produce the structured output format | Structured output enables comparison, audit trails, and handoff to other teams |
Output Format
Produce a structured report with:
- Cost baseline with trends
- Waste inventory (idle/unused resources with monthly cost)
- Rightsizing candidates with specific recommendations
- Commitment opportunities with ROI calculations
- Prioritized action plan sorted by savings impact
- Executive summary for leadership sharing