TheDocumentation Index
Fetch the complete documentation index at: https://mintlify.com/aws-samples/sample-well-architected-skills-and-steering/llms.txt
Use this file to discover all available pages before exploring further.
sustainability-optimization skill teaches your AI coding agent to assess workloads against the AWS Well-Architected Sustainability pillar. It examines actual code and infrastructure-as-code to find resource waste, inefficient compute choices, missing data lifecycle policies, and architectural patterns that consume more energy than necessary — then produces a prioritized remediation plan with concrete efficiency gains.
What it does
Compute Efficiency
Identifies x86 workloads where Graviton/arm64 applies, always-on resources that could scale to zero, and non-production environments running 24/7 at full capacity.
Data & Storage Lifecycle
Flags S3 buckets without lifecycle policies, CloudWatch logs with unlimited retention, DynamoDB tables missing TTL, and uncompressed data stores.
Architecture Patterns
Finds polling patterns that could be event-driven, chatty APIs causing excess network round-trips, and self-managed infrastructure where managed equivalents offer better utilization.
Development Practices
Reviews Docker image sizes, CI/CD caching, test environment teardown policies, and artifact retention to reduce build and delivery waste.
WA Sustainability pillar coverage
The skill evaluates all six Sustainability pillar questions with evidence sourced from your codebase.SUS 1 — Region selection
SUS 1 — Region selection
Reviews region configurations and documentation for carbon-intensity rationale. Flags deployments in high-carbon regions where latency requirements don’t justify the placement.
SUS 2 — User behavior patterns
SUS 2 — User behavior patterns
Examines auto-scaling configurations, scheduled scaling rules, and event-driven architecture patterns to verify resources can scale to zero during idle periods.
SUS 3 — Software and architecture patterns
SUS 3 — Software and architecture patterns
Checks for managed service adoption, asynchronous processing, batch operations, and caching layers that eliminate redundant computation.
SUS 4 — Data access and usage patterns
SUS 4 — Data access and usage patterns
Audits S3 lifecycle policies, Intelligent-Tiering adoption, compression configurations, TTL settings on DynamoDB tables, and backup retention policies.
SUS 5 — Hardware management and usage
SUS 5 — Hardware management and usage
Reports Graviton/arm64 adoption rate across Lambda, EC2, Fargate, and ECS. Flags older instance generations (m4, t2) and provisioned concurrency on non-latency-critical Lambdas.
SUS 6 — Development and deployment processes
SUS 6 — Development and deployment processes
Reviews multi-stage Docker builds, CI/CD layer caching, incremental build configurations, and test environment auto-teardown policies.
How to invoke it
Ask your AI coding agent any of the following — the skill activates automatically:What the agent analyzes
Compute efficiency discovery
The agent scans instance type selections, Lambda architecture settings (
arm64 vs x86_64), auto-scaling min/max values, and whether provisioned concurrency is justified. It documents every compute resource with its file path and line number.Data and storage lifecycle
Every S3 bucket, DynamoDB table, CloudWatch log group, and backup policy is inspected. Missing lifecycle rules, unlimited retention, and absent TTL configurations are flagged with direct code references.
Architecture efficiency
The agent identifies polling loops, synchronous chains that could be async, individual API calls where batch operations exist, and self-managed infrastructure (Kafka, Redis) where managed equivalents (MSK, ElastiCache) provide shared-infrastructure efficiency.
Checkpoint — confirm before proceeding
The skill pauses and summarizes its discoveries before evaluating findings against the WA framework questions. You confirm before the assessment continues.
WA framework evaluation
Each of SUS 1–6 is scored with status, evidence (file:line), gaps, and risk level.
Example output
The skill produces a structured Sustainability Assessment report:Resource efficiency opportunities
The report includes a dedicated table mapping current configurations to optimized alternatives:Effectiveness
Evaluated using an automated LLM-as-judge framework with paired comparison (same prompt, with and without skill context) using Claude Opus 4.8.
| Baseline | With skill | Delta | |
|---|---|---|---|
| Score | 85% | 100% | +15% |
AWS tools referenced in the output
Customer Carbon Footprint Tool
Track your AWS emissions directly in the AWS Management Console — the skill recommends enabling this to establish a baseline.
AWS Compute Optimizer
Right-size recommendations with explicit Graviton suggestions. The skill flags which resources should be submitted to Compute Optimizer for runtime analysis.
S3 Storage Lens
Storage efficiency insights and access pattern analysis — recommended for buckets where Intelligent-Tiering decision cannot be made from code alone.
AWS Trusted Advisor
Idle resource detection for runtime waste not visible in IaC. The skill notes when Trusted Advisor data would sharpen a finding.
Related skills
| Skill | When to use instead |
|---|---|
cost-optimization-review | Focus on spend reduction, right-sizing, and pricing model changes rather than carbon footprint |
performance-efficiency | Focus on latency, throughput, and resource selection for performance targets |
wa-review | Run a full cross-pillar review including Sustainability alongside all five other pillars |