Skip to main content

Documentation 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.

Skills are self-contained, tool-agnostic playbooks that any AI agent can follow step-by-step. They don’t depend on steering files or on each other — each skill is a complete, standalone guide that embeds AWS Well-Architected Framework best practices directly into your development workflow. Drop one into any supported AI coding tool and invoke it with a natural-language prompt.

​
All 11 Skills

The table below maps each skill to the Well-Architected pillar(s) it covers and the scenario it’s designed for.
SkillPillar(s)Use when you need to…
wa-reviewAll 6Run a full Well-Architected review
security-assessmentSecurityAssess IAM, detection, data protection, incident response
reliability-improvement-planReliabilityFind and eliminate single points of failure
cost-optimization-reviewCost OptimizationIdentify waste and right-sizing opportunities
performance-efficiencyPerformance EfficiencyEvaluate resource selection, scaling, and caching
sustainability-optimizationSustainabilityReduce carbon footprint and resource waste
operational-excellenceOperational ExcellenceAssess CI/CD, observability, incident management
migration-readinessAll 6Assess readiness to migrate a workload to AWS
architecture-decision-recordAll 6Document a design decision with WA pillar impact
wa-builderAll 6Learn WA for your project + produce visual artifacts (diagrams, decision trees, roadmaps)
wa-guardrailsAll 6Generate preventive controls (Config rules, SCPs, CI checks, alarms) for ongoing adherence to WA guidance

​
Browse by Skill

WA Review

Full 57-question Well-Architected review across all 6 pillars. Evidence-backed findings with Eisenhower-prioritized remediation.

Security Assessment

Deep-dive into IAM, encryption, network protection, detection controls, and incident response preparedness.

Reliability Improvement Plan

Identify single points of failure, assess recovery capabilities, and improve change management safety.

Cost Optimization Review

Find waste, right-sizing opportunities, and pricing model mismatches across your AWS workload.

Performance Efficiency

Evaluate resource selection, caching strategies, scaling configurations, and latency monitoring.

Sustainability Optimization

Reduce carbon footprint by improving compute efficiency, data lifecycle, and managed service adoption.

Operational Excellence

Assess CI/CD pipelines, observability, incident management, and operational automation.

Migration Readiness

7 Rs assessment with a full migration plan covering all 6 Well-Architected pillars.

Architecture Decision Record

Document design decisions with explicit Well-Architected pillar impact analysis.

WA Builder

Learn Well-Architected principles for your project and produce visual artifacts — diagrams, decision trees, roadmaps.

WA Guardrails

Generate preventive controls: AWS Config rules, SCPs, CI/CD checks, and CloudWatch alarms.

​
How to Invoke Skills

Skills are invoked with natural-language prompts. Each tool exposes them differently — some use slash commands, others load them as rules or instructions that the agent applies automatically.
Claude Code exposes skills as slash commands. After installing the adapter, invoke any skill directly from the chat input:
/wa-review
/security-assessment
/reliability-improvement-plan
/cost-optimization-review
/performance-efficiency
You can also use natural-language prompts in any conversation:
Run a WA review on this codebase
Do a security review — focus on IAM and encryption
Find single points of failure in our infrastructure
Check our cost optimization posture

​
Skill vs Steering

Skills and steering serve different purposes. Skills are explicit playbooks — you invoke them for a specific task and the agent follows a structured, multi-step process. Steering is always-on context that shapes how the agent reasons about architecture in every conversation, without an explicit invocation. Most tools use both together for the best results.
SkillsSteering
When activeWhen explicitly invokedEvery conversation
ScopeOne focused taskContinuous background guidance
DepthDeep, structured, step-by-stepLightweight, principle-level
OutputFormal report or planInline suggestions and corrections
ToolsAll 13 supported toolsKiro (native), others via adapters

​
Effectiveness

All skills are evaluated using an automated LLM-as-judge framework comparing the same prompt with and without skill context. Results with Claude Opus 4.8 (generation and grading), 16K token output:
SkillBaselineWith SkillDelta
wa-review82%100%+18%
architecture-decision-record81%100%+19%
cost-optimization-review93%100%+7%
migration-readiness85%100%+15%
operational-excellence90%100%+10%
performance-efficiency90%100%+10%
reliability-improvement-plan95%100%+5%
security-assessment94%100%+6%
sustainability-optimization85%100%+15%
wa-builder61%94%+33%
wa-guardrails76%99%+23%
Average85%99%+15%
9 of 11 skills score 100% on behavioral assertions. The average improvement is +15% over the same model without skill guidance, and no skill ever produces worse output than the baseline.

Build docs developers (and LLMs) love

Get started for freeTalk to us