[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7A3EN774JBJB1WCF8KWD3G7":3,"skill-same:01M7A3EN774JBJB1WCF8KWD3G7":149},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":45,"commitSha":46,"skillDirectory":47,"forksCount":48,"repositoryHtmlUrl":49,"repositoryHomepageUrl":6,"ownerHtmlUrl":50,"ownerBlogUrl":6,"ownerDescription":51,"updatedAt":52,"install":53,"resources":66},"01M7A3EN774JBJB1WCF8KWD3G7","apprunner-autoscaling-optimizer",null,"Optimises AWS App Runner autoscaling configuration for cost and performance — auto-scaling configuration (MinSize\u002FMaxSize provisioning), concurrency setting (requests per instance, the primary cost lever), CPU\u002Fmemory utilization target tuning, instance type sizing (1 vCPU \u002F 2 GB vs 2 vCPU \u002F 4 GB vs 4 vCPU \u002F 8 GB), health check interval tuning, scale-in cooldown, static vs dynamic traffic pattern analysis, cost-per-request modeling, VPC ingress\u002F egress cost analysis (NAT gateway, VPC endpoints), observability cost, and pause\u002Fresume for non-prod cost savings. Reads App Runner service configuration, CloudWatch metrics, and Cost Explorer data to project monthly savings. Emits OPTIMIZED with specific recommendation and estimated savings, or FURTHER_OPTIMIZATION_AVAILABLE. Use when reviewing App Runner spend, tuning concurrency, right-sizing MinSize, or evaluating pause\u002Fresume for non-prod environments.","jhkchan\u002Faws-agent-skills","jhkchan","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1083902?v=4","01M7A2SVQYP9GQTPTDXP5JKYQ4","Jacky Chan","01M7A2SVS117YWCTZABB75HZR1",4,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: apprunner-autoscaling-optimizer\ndescription: Optimises AWS App Runner autoscaling configuration for cost and performance — auto-scaling configuration (MinSize\u002FMaxSize provisioning), concurrency setting (requests per instance, the primary cost lever), CPU\u002Fmemory utilization target tuning, instance type sizing (1 vCPU \u002F 2 GB vs 2 vCPU \u002F 4 GB vs 4 vCPU \u002F 8 GB), health check interval tuning, scale-in cooldown, static vs dynamic traffic pattern analysis, cost-per-request modeling, VPC ingress\u002F egress cost analysis (NAT gateway, VPC endpoints), observability cost, and pause\u002Fresume for non-prod cost savings. Reads App Runner service configuration, CloudWatch metrics, and Cost Explorer data to project monthly savings. Emits OPTIMIZED with specific recommendation and estimated savings, or FURTHER_OPTIMIZATION_AVAILABLE. Use when reviewing App Runner spend, tuning concurrency, right-sizing MinSize, or evaluating pause\u002Fresume for non-prod environments.\nlicense: Apache-2.0\ncompatibility: Agent runtime that reads SKILL.md (Claude Code, Cursor, Windsurf, Codex, Gemini). Offline recommendation classification works from pasted App Runner service configuration and CloudWatch metrics. Live-account optimization uses aws apprunner describe-service, describe-auto-scaling- configuration, list-services, list-auto-scaling-configurations, aws cloudwatch get-metric-statistics (RequestCount, InstanceCount, CPUUtilization, MemoryUtilization, 4xx, 5xx, Latency), aws ce get-cost- and-usage...\nmetadata:\n  domain: aws-cloudops\n  complexity: medium\n  requires_llm: 'true'\n  phase: '3'\n  supports_pipeline: 'true'\n  entry_point: 'false'\n  family: Compute\n  task_type: optimize\n  skill_class: capability\n  lifecycle_status: active\n  verdict_shape: OPTIMIZED | FURTHER_OPTIMIZATION_AVAILABLE\n  when_to_use: Optimising App Runner autoscaling configuration, tuning concurrency settings, right-sizing MinSize\u002FMaxSize, evaluating instance type upgrades (1\u002F2\u002F4 vCPU), analyzing cost-per-request, deciding whether to pause non-prod services, reviewing VPC egress cost, tuning health check intervals, or conducting a FinOps review of App Runner services.\n  when_not_to_use: ECS\u002FFargate service autoscaling (use an ECS autoscaling skill), Lambda cost optimization (use lambda-cost-optimizer), EC2 instance rightsizing (use ec2-rightsizing-optimizer), or App Runner troubleshooting (deployment failures, configuration errors — use the App Runner troubleshooter). This skill focuses on cost-driven autoscaling optimization, not functional debugging.\n  activation_triggers: optimise App Runner, App Runner autoscaling, App Runner concurrency, App Runner MinSize, App Runner MaxSize, App Runner instance type, App Runner cost, App Runner FinOps, App Runner pause, App Runner resume, App Runner VPC egress, App Runner health check, App Runner scale-in cooldown, App Runner cost per request, App Runner right-size, App Runner provisioned, reduce App Runner bill, App Runner non-prod savings\n  invocation_schema: 'Input: either (a) an App Runner service ARN + live-account context, (b) a service configuration document (describe-service + describe-auto-scaling-configuration output), OR (c) CloudWatch metrics (RequestCount, InstanceCount, CPUUtilization, MemoryUtilization, 4xx, 5xx) with at least 14 days of observation. Output: a deterministic TARGET\u002FVERDICT\u002FREASON\u002FRECOMMENDATION\u002FESTIMATED_SAVINGS\u002FMIGRATION_STEPS block per service, where VERDICT is one of OPTIMIZED, FURTHER_OPTIMIZATION_AVAILABLE.'\n  invocation_example: \"# Minimal valid input (offline finding classification): ServiceName: prod-api-service AutoScalingConfiguration:\\n  MinSize: 1\\n  MaxSize: 25\\n  Concurrency: 100\\nInstanceConfiguration:\\n  Cpu: 1 vCPU\\n  Memory: 2 GB\\nRegion: us-east-1 Metrics (last 30 days):\\n  - RequestCount: 8,000,000\u002Fmonth\\n  - Avg InstanceCount: 2.3 (MinSize 1 is sufficient)\\n  - Avg CPUUtilization: 15%\\n  - Avg MemoryUtilization: 22%\\n  - 5xx rate: 0.01%\\nTraffic pattern: dynamic (peak 3x baseline during business hours) Cost (last 30 days): $1,200\u002Fmonth Emit the standard optimization block (TARGET, VERDICT, REASON, RECOMMENDATION, ESTIMATED_SAVINGS, MIGRATION_STEPS).\"\n  version: 0.1.0\n  author: Jacky Chan — AWS Community Builder\n  keywords: App Runner, auto-scaling configuration, concurrency, MinSize, MaxSize, provisioned instances, instance type, 1 vCPU 2 GB, 2 vCPU 4 GB, 4 vCPU 8 GB, CPU utilization, memory utilization, health check, health check interval, scale-in cooldown, scale-out speed, static traffic, dynamic traffic, cost per request, custom domain, SSL certificate, deployment, VPC ingress, VPC egress, NAT gateway, observability cost, CloudWatch Logs, pause service, resume service, non-prod savings, FinOps\n  tags: aws, apprunner, compute, cost-optimization, autoscaling, finops, concurrency, serverless-container, optimize\n---\n\n# App Runner Autoscaling Optimizer\n\n## What this skill does\n\nTranslates an App Runner service's autoscaling posture into a concrete\ncost-optimization recommendation with a dollar-denominated savings\nestimate. The verdict is the highest-leverage action across six dimensions\n— concurrency tuning, MinSize\u002FMaxSize right-sizing, instance type sizing,\nhealth check optimization, VPC egress cost reduction, and pause\u002Fresume for\nnon-prod — applied in priority order. Always pairs the recommendation with\nexact CLI commands and a projected monthly savings figure.\n\n## Quick navigation\n\n| Section | What it covers | When to read it |\n|---|---|---|\n| **Quick reference** | Verdict thresholds (OPTIMIZED\u002FFURTHER_OPTIMIZATION_AVAILABLE) + optimization priority | Before any recommendation |\n| **Mindset** | Why concurrency is the primary lever, the MinSize baseline tradeoff, pause-for-savings | Understanding the cost model |\n| **Pre-flight** | Service metadata gate — auto-scaling config, instance config, traffic pattern | Before emitting any recommendation |\n| **Process** | Per-dimension analysis: concurrency, MinSize, instance type, health check, VPC, pause\u002Fresume | When choosing which optimization to apply |\n| **Output format** | Structured TARGET\u002FVERDICT\u002FREASON\u002FRECOMMENDATION\u002FSAVINGS template | Formatting the response |\n| **Anti-Patterns** | NEVER list — common mistakes that break scaling or increase cost | Review before risky changes |\n\n## Quick reference — verdict thresholds\n\n| Verdict | Trigger condition | Action |\n|---|---|---|\n| `OPTIMIZED` | All optimization dimensions checked; service configuration matches traffic pattern with no actionable savings > 5% | Emit confirmation, monitoring plan |\n| `FURTHER_OPTIMIZATION_AVAILABLE` | At least one dimension has actionable savings >= 5% or a structural change (pause\u002Fresume, instance type, concurrency) is recommended | Emit recommendation, CLI steps, estimated savings |\n\n**Optimization priority (apply in this order, each may yield independent\nsavings):**\n\n1. **Pause\u002Fresume for non-prod** — if the service is dev\u002Fstaging\u002FQA and\n   runs 24\u002F7, pausing outside business hours saves 100% of compute cost\n   (~70% of total App Runner bill). Highest single-action savings.\n2. **Concurrency tuning** — the primary lever. Higher concurrency = fewer\n   instances for the same traffic = lower cost. But higher concurrency\n   increases per-instance latency and memory pressure. Find the sweet spot.\n3. **MinSize right-sizing** — MinSize = always-provisioned baseline. For\n   static traffic, set MinSize to the steady-state instance count. For\n   dynamic traffic, set MinSize to 0 (scale to zero) or 1 (always-on).\n4. **Instance type sizing** — 1 vCPU\u002F2 GB is the cheapest; 2 vCPU\u002F4 GB\n   and 4 vCPU\u002F8 GB are 2x and 4x the cost. If CPU\u002Fmemory utilization is\n   low, the instance type may be overprovisioned. If concurrency is high\n   and latency is elevated, a larger instance type with higher concurrency\n   may be more cost-efficient.\n5. **Health check optimization** — longer intervals reduce false-positive\n   restarts but delay detection of unhealthy instances. Default 5s interval\n   is aggressive for most workloads.\n6. **VPC egress cost** — if the service uses a VPC connector, NAT gateway\n   charges apply to all egress. A VPC endpoint for AWS services (S3, DynamoDB,\n   etc.) eliminates NAT gateway cost for that traffic.\n\n**Cost baselines (2026, us-east-1):**\n\n| Instance type | Compute (per hour, provisioned) | Compute (per second, on-demand) | Best for |\n|---|---|---|---|\n| 1 vCPU \u002F 2 GB | $0.063\u002Fhr | $0.0000175\u002Fs | Low-traffic, small workloads |\n| 2 vCPU \u002F 4 GB | $0.126\u002Fhr | $0.0000350\u002Fs | Standard web APIs |\n| 4 vCPU \u002F 8 GB | $0.252\u002Fhr | $0.0000700\u002Fs | CPU\u002Fmemory-intensive workloads |\n\n- On-demand billing: per-second charges only when processing requests\n  (scale-to-zero eliminates idle cost).\n- Provisioned billing (MinSize >= 1): per-hour charges for the baseline\n  instances even when idle.\n- Concurrency does not have a direct charge — it controls how many\n  concurrent requests one instance handles, which indirectly controls\n  how many instances the autoscaler provisions.\n\n## Mindset\n\n**One-line takeaway:** Concurrency is the primary cost lever. Higher\nconcurrency means each instance handles more simultaneous requests, so\nthe autoscaler provisions fewer instances — but at the cost of increased\nper-request latency and memory pressure. Find the sweet spot where\nconcurrency is high enough to minimize instance count, but low enough to\nkeep latency within the SLO. Driven by three App Runner realities:\n\n- **Concurrency determines instance count.** App Runner scales based on\n  concurrent requests. With 100 concurrent requests and a concurrency\n  setting of 50, the autoscaler provisions 2 instances. With a concurrency\n  setting of 100, it provisions 1 instance. The cost difference is ~50%.\n  The latency difference depends on the workload — a CPU-bound request at\n  concurrency 100 takes longer than at concurrency 50 because the instance\n  is time-slicing.\n\n- **MinSize is the cost floor.** A MinSize of 1 means one instance is\n  always provisioned and billed at the hourly rate, even at 3am with zero\n  traffic. For dynamic traffic (web APIs with business-hours peaks), MinSize\n  0 (scale-to-zero) eliminates the idle cost. For static traffic (always-on\n  services), MinSize should match the steady-state instance count so the\n  autoscaler does not over-provision.\n\n- **Pause\u002Fresume for non-prod saves 100%.** App Runner services can be\n  paused (not deleted). A paused service incurs zero compute charges. For\n  dev\u002Fstaging\u002FQA environments that are only used during business hours,\n  pausing outside those hours saves ~75% of the monthly compute cost\n  (18 hours paused \u002F 24 hours). Combined with scale-to-zero during business\n  hours, non-prod services can run at a fraction of production cost.\n\n## Pre-flight: service metadata gate\n\nRun before classification. Misclassifying these produces wrong\nrecommendations.\n\n**Live-account pre-flight (skip if offline plan audit):**\nLive-account pre-flight command listing moved verbatim to [references\u002Fdiagnostic-commands.md](references\u002Fdiagnostic-commands.md).\nLoad on demand before a live-account optimization; offline plan audits skip it.\n\n**Malformed input:** if the input JSON is invalid or missing required\nfields, emit `VERDICT: ERROR` with `REASON: Service configuration is not\nvalid JSON or is missing required fields — cannot plan.` and\n`REMEDIATION: Re-fetch with aws apprunner describe-service --service-arn\n\u003Carn> --output json and re-plan.`\n\n| Service attribute | Effect on optimization |\n|---|---|\n| `Status: PAUSED` | No compute charges. Resume-related recommendation only. |\n| `MinSize: 0` (scale-to-zero) | Already optimized for idle cost. Check if cold-start latency is acceptable. |\n| `MinSize >= 1` | Baseline compute cost. Check if MinSize matches steady-state instance count. |\n| `Concurrency: 1` (minimum) | Maximum instance count = maximum cost. Almost always overprovisioned. |\n| `Concurrency: 1000` (maximum) | Minimum instance count = minimum cost. Check if latency is within SLO. |\n| `CPUUtilization \u003C 20%` (sustained) | Instance type may be overprovisioned. |\n| `CPUUtilization > 70%` (sustained) | Instance type may be underprovisioned, or concurrency too high. |\n| `MemoryUtilization > 80%` (sustained) | Concurrency too high or instance memory too low. Risk of OOM. |\n| `5xxResponseCount` spikes | Possible OOM from high concurrency or slow health checks. |\n| VPC connector with NAT gateway | Egress cost on all outbound traffic. |\n| `HealthCheckConfiguration.Interval: 1` | Aggressive health checking. May cause false-positive restarts. |\n| Custom domain with managed certificate | Negligible compute overhead. No optimization needed. |\n\n## Process — optimization analysis (apply in priority order)\n\n### Step 0: Expert knowledge — non-obvious App Runner autoscaling behaviors\n\nStep 0 expert-knowledge behaviors moved verbatim to [references\u002Fadvanced-patterns.md](references\u002Fadvanced-patterns.md).\nLoad on demand when a recommendation depends on non-obvious behavior (HPA stabilization window, non-linear concurrency\u002FCPU curve, resume cold-pull).\n\n### Step 1: Pause\u002Fresume analysis (non-prod only)\n\nIf the service is in a non-prod environment (dev\u002Fstaging\u002FQA):\n\n```\nIf service runs 24\u002F7:\n  Recommend pause outside business hours (e.g., 7pm-7am + weekends)\n  Savings = monthly compute * (paused_hours \u002F total_hours)\n  Typical: 18hr\u002Fday paused on weekdays + 48hr weekends = ~75% savings\n\nIf service is already paused or uses MinSize 0:\n  No pause\u002Fresume recommendation. Proceed to concurrency analysis.\n```\n\n**CLI for pause\u002Fresume automation:**\n\n```bash\n# Pause (using EventBridge scheduled rule)\naws apprunner pause-service --service-arn \u003Carn>\n\n# Resume\naws apprunner resume-service --service-arn \u003Carn>\n```\n\n### Step 2: Concurrency analysis\n\n```\ncurrent_concurrency = AutoScalingConfiguration.Concurrency\navg_concurrent_requests = max(RequestCount_per_5min \u002F 300) (peak concurrent)\navg_instance_count = average(InstanceCount over 30 days)\nobserved_concurrency_per_instance = avg_concurrent_requests \u002F avg_instance_count\n\nIf observed_concurrency_per_instance \u003C current_concurrency * 0.5:\n  → Concurrency is overprovisioned (instances are underutilized)\n  → Recommend: increase concurrency to reduce instance count\n  → New concurrency = current_concurrency * 1.5 (gradual increase)\n  → Savings = (current_instances - projected_instances) * hourly_rate * 730\n\nIf observed_concurrency_per_instance > current_concurrency * 0.8:\n  AND CPUUtilization > 60% OR MemoryUtilization > 75%:\n  → Concurrency is near the limit; increasing further risks latency\u002FOOM\n  → Recommend: increase instance type (more CPU\u002Fmemory per instance)\n  → Then increase concurrency to match\n\nIf latency p95 > SLO target:\n  → Concurrency may be too high for the workload\n  → Recommend: decrease concurrency to reduce per-request latency\n  → Cost increases but stays within SLO\n```\n\n### Step 3: MinSize\u002FMaxSize right-sizing\n\n```\nIf MinSize > 0:\n  steady_state_instances = percentile(InstanceCount, 10) (low-traffic baseline)\n  If MinSize > steady_state_instances:\n    → MinSize is overprovisioned; reduce to steady_state_instances\n    → If steady_state_instances = 0: recommend MinSize 0 (scale-to-zero)\n\nIf MaxSize \u003C peak(InstanceCount):\n  → MaxSize is too low; instances are hitting the cap during peaks\n  → Recommend: increase MaxSize to peak(InstanceCount) * 1.5\n\nIf MaxSize > peak(InstanceCount) * 3:\n  → MaxSize is excessively high; no immediate cost impact but risk of\n    runaway scaling\n  → Recommend: reduce MaxSize to peak(InstanceCount) * 2\n```\n\n### Step 4: Instance type analysis\n\n```\ncpu_avg = average(CPUUtilization over 30 days)\nmem_avg = average(MemoryUtilization over 30 days)\n\nIf cpu_avg \u003C 25% AND mem_avg \u003C 40%:\n  AND current instance type > 1 vCPU \u002F 2 GB:\n  → Instance type is overprovisioned\n  → Recommend: downgrade to 1 vCPU \u002F 2 GB\n  → Savings = (current_hourly - new_hourly) * instance_count * 730\n\nIf cpu_avg > 70% OR mem_avg > 80%:\n  → Instance type is underprovisioned\n  → Recommend: upgrade to next size\n  → Also consider: increasing concurrency with larger instance type\n```\n\n### Step 5: Health check optimization\n\n```\nIf HealthCheckConfiguration.Interval \u003C 5:\n  AND InstanceRestartCount is high:\n  → Health check interval is too aggressive; false-positive restarts\n  → Recommend: increase interval to 10-15s\n\nIf HealthCheckConfiguration.Timeout > Interval * 0.5:\n  → Timeout is too close to interval; health checks may always timeout\n  → Recommend: set timeout to Interval * 0.3\n\nIf HealthCheckConfiguration.UnhealthyThreshold \u003C 3:\n  → Single failed check triggers replacement; too sensitive\n  → Recommend: set to 3-5\n```\n\n### Step 6: VPC egress cost analysis\n\n```\nIf NetworkConfiguration.EgressConfiguration uses NAT gateway:\n  nat_cost = CostExplorer filter \"NatGateway\" for the VPC\n  If nat_cost > compute_cost * 0.3:\n    → NAT gateway cost is significant\n    → Recommend: add VPC endpoints for AWS service traffic\n    → Savings = nat_cost * (aws_service_traffic_fraction)\n```\n\n### Step 7: Synthesize — emit verdict\n\nIf any dimension yielded actionable savings >= 5%:\n→ `VERDICT: FURTHER_OPTIMIZATION_AVAILABLE`\n\nIf all dimensions are optimal (no actionable savings >= 5%):\n→ `VERDICT: OPTIMIZED`\n\n## Output format (per service)\n\n```text\nTARGET: \u003Cservice-name> (region: \u003Cregion>)\nVERDICT: OPTIMIZED | FURTHER_OPTIMIZATION_AVAILABLE\nREASON: \u003Cone-line summary of the primary finding>\nRECOMMENDATION:\n  - \u003Cspecific recommendation 1>\n  - \u003Cspecific recommendation 2>\nESTIMATED_SAVINGS: $\u003Cmonthly_savings>\u002Fmonth (\u003Cpercentage>% of current spend)\nMIGRATION_STEPS:\n  1. \u003CCLI command or action>\n  2. \u003CCLI command or action>\n  3. \u003Cverification step>\nCURRENT_MONTHLY_COST: $\u003Camount>\nPROJECTED_MONTHLY_COST: $\u003Camount>\nTRAFFIC_PATTERN: \u003Cstatic | dynamic | bursty | intermittent>\nRISK: \u003CLOW | MODERATE | HIGH>\nNOTES: \u003Ccaveats, monitoring recommendations>\n```\n\n### Worked example — concurrency tuning (FURTHER_OPTIMIZATION_AVAILABLE)\n\n```text\nTARGET: prod-api-service (region: us-east-1)\nVERDICT: FURTHER_OPTIMIZATION_AVAILABLE\nREASON: Concurrency 100 is overprovisioned — observed peak is 42 concurrent\n        requests per instance, meaning 58% of provisioned capacity is unused.\nRECOMMENDATION:\n  - Increase concurrency from 100 to 150 (gradual 50% increase)\n  - Current avg 2.3 instances → projected 1.5 instances at concurrency 150\n  - Monitor latency p95 for 7 days post-change\nESTIMATED_SAVINGS: $230\u002Fmonth (19% of current compute spend)\nMIGRATION_STEPS:\n  1. Create new auto-scaling configuration:\n     aws apprunner create-auto-scaling-configuration \\\n       --auto-scaling-configuration-name prod-api-service-asg-v2 \\\n       --min-size 1 \\\n       --max-size 10 \\\n       --concurrency 150\n  2. Apply to the service:\n     aws apprunner update-service \\\n       --service-arn arn:aws:apprunner:us-east-1:111111111111:service\u002Fprod-api-service\u002Fabc \\\n       --auto-scaling-configuration-arn arn:aws:apprunner:us-east-1:111111111111:autoscalingconfiguration\u002Fprod-api-service-asg-v2\u002F1\n  3. Monitor for 7 days:\n     aws cloudwatch get-metric-statistics \\\n       --namespace AWS\u002FAppRunner \\\n       --metric-name Latency \\\n       --dimensions Name=ServiceId,Value=\u003Cid> \\\n       --statistics Average \\\n       --period 300 --start-time ... --end-time ...\n  4. If latency p95 increases > 20%, revert to concurrency 100\nCURRENT_MONTHLY_COST: $1,200\u002Fmonth\nPROJECTED_MONTHLY_COST: $970\u002Fmonth\nTRAFFIC_PATTERN: dynamic (3x peak during business hours)\nRISK: MODERATE\nNOTES:\n  - Concurrency 150 is still below the workload's CPU limit (avg\n    CPUUtilization at concurrency 100 is 35%; at 150 it is projected\n    ~52%).\n  - If latency degrades, consider upgrading to 2 vCPU \u002F 4 GB instead\n    of reducing concurrency back.\n```\n\n### Worked example — pause\u002Fresume for non-prod (FURTHER_OPTIMIZATION_AVAILABLE)\n\nSecondary worked example moved verbatim to [references\u002Fworked-examples.md](references\u002Fworked-examples.md).\nLoad on demand when emitting a pause\u002Fresume recommendation.\n\n### Worked example — already optimal (OPTIMIZED)\n\nSecondary worked example moved verbatim to [references\u002Fworked-examples.md](references\u002Fworked-examples.md).\nLoad on demand when confirming a service is already optimal.\n\n## Anti-Patterns — NEVER\n\n- NEVER increase concurrency without monitoring latency for at least 7 days\n  post-change. Higher concurrency reduces instance count but increases per-\n  request latency on CPU-bound workloads. A blind increase from 50 to 200\n  can degrade p95 latency by 3-5x.\n\n- NEVER set MinSize to 0 for a production service that requires low cold-\n  start latency. App Runner cold start after scale-to-zero is 5-30 seconds\n  (container pull + start). For user-facing APIs, keep MinSize at 1.\n\n- NEVER pause a production service. Pause is for non-prod only. Pausing\n  removes the endpoint and takes the service offline.\n\n- NEVER downgrade instance type based on a single day of low utilization.\n  CPU\u002Fmemory utilization varies by day of week and time of day. Use at\n  least 14 days of data, ideally 30, to capture the full traffic cycle.\n\n- NEVER set MaxSize below the peak observed instance count. The autoscaler\n  will hit the cap and return 5xx errors during traffic spikes.\n\n- NEVER ignore VPC egress cost. For services with a VPC connector, NAT\n  gateway charges ($0.045\u002FGB processed + $0.045\u002FGB data transfer) can\n  exceed compute cost for high-egress workloads. Use VPC endpoints for\n  AWS service traffic.\n\n- NEVER assume App Runner is the right platform for all workloads. For\n  sustained high-traffic services (> $2,000\u002Fmonth), ECS\u002FFargate or EC2 may\n  be more cost-efficient. Run a cost-per-request comparison.\n\n- NEVER change concurrency, MinSize, and instance type simultaneously. Make\n  one change at a time and measure the impact for 7 days before the next.\n  Simultaneous changes make it impossible to attribute cost\u002Flatency\n  changes to a specific dimension.\n\n- NEVER delete an auto-scaling configuration that is still referenced by a\n  service. `delete-auto-scaling-configuration` fails if the configuration\n  is in use. Use `update-service` to switch to the new configuration first.\n\n- NEVER set health check interval to 1 second unless the workload requires\n  instant failure detection. A 1-second interval with a 3-second timeout\n  and 3 unhealthy threshold means 3 seconds to detect a failure — but it\n  also means transient latency spikes (a slow request holding the health\n  check thread) trigger false-positive restarts. Use 5-10 seconds.\n\n- NEVER reduce concurrency to reduce latency without checking if the\n  workload is CPU-bound or I\u002FO-bound. For I\u002FO-bound workloads (API calls,\n  database queries), concurrency has minimal latency impact — the instance\n  handles concurrent I\u002FO natively. For CPU-bound workloads, concurrency\n  directly increases latency.\n\n- NEVER assume scale-to-zero eliminates all cost. App Runner charges for\n  container image storage in ECR and any custom domain certificate\n  management. These are typically negligible (\u003C $5\u002Fmonth) but not zero.\n\n- NEVER forget that deployments pause autoscaling. During a deployment, the\n  autoscaler does not add or remove instances. High-traffic deployments may\n  cause temporary latency spikes. Schedule deployments during low-traffic\n  windows.\n\n- NEVER auto-execute a state-changing App Runner CLI without the CONFIRM\n  gate. Changing the auto-scaling configuration affects production traffic.\n  The service briefly transitions during configuration changes.\n\n## Pre-flight safety checks (run before any optimization CLI)\n\nPre-flight safety checks (CONFIRMATION GATE, pre-state capture, post-change monitoring) moved verbatim to [references\u002Fdiagnostic-commands.md](references\u002Fdiagnostic-commands.md).\nLoad on demand before executing any state-changing optimization CLI.\n\n## Recent AWS features (2024-2026)\n\nRecent AWS features (2024-2026) moved verbatim to [references\u002Fadvanced-patterns.md](references\u002Fadvanced-patterns.md).\nLoad on demand when evaluating v2 native VPC networking, blue\u002Fgreen deployments, or the observability dashboard.\n\n## References (load on demand)\n\n- [references\u002Fworked-examples.md](references\u002Fworked-examples.md) — secondary worked examples moved from SKILL.md: pause\u002Fresume for non-prod, already-optimal OPTIMIZED output.\n- [references\u002Fadvanced-patterns.md](references\u002Fadvanced-patterns.md) — Step 0 non-obvious autoscaling behaviors and Recent AWS features (2024-2026) moved from SKILL.md.\n- [references\u002Fdiagnostic-commands.md](references\u002Fdiagnostic-commands.md) — live-account pre-flight command listing and pre-flight safety checks moved from SKILL.md.\n- [references\u002Fapprunner-pricing-and-sizing.md](references\u002Fapprunner-pricing-and-sizing.md) — pricing model, instance-type pricing, monthly cost examples, sizing heuristics.\n- [references\u002Foptimization-procedures.md](references\u002Foptimization-procedures.md) — per-dimension optimization procedures (concurrency, MinSize, instance type, health check, VPC, pause\u002Fresume).\n\n## Domain\n\nAWS CloudOps \u002F App Runner Autoscaling Cost Optimization & FinOps.\n\n## AWS documentation\n\n- **AWS App Runner Developer Guide** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fwhat-is-apprunner.html\n- **App Runner auto-scaling** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-autoscaling.html\n- **App Runner autoscaling configuration** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-autoscaling.html#manage-autoscaling.config\n- **App Runner instance configuration** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-configure.html\n- **App Runner VPC networking** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fnetwork-vpc.html\n- **App Runner health checks** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-configure.html#manage-configure.health\n- **App Runner API Reference** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fapi\u002FWelcome.html\n- **App Runner pricing** — https:\u002F\u002Faws.amazon.com\u002Fapprunner\u002Fpricing\u002F\n- **App Runner pause\u002Fresume** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-delete.html#manage-delete.pause\n- **CloudWatch App Runner metrics** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmonitor-cw.html\n","# App Runner Autoscaling Optimizer\n\n## What this skill does\n\nTranslates an App Runner service's autoscaling posture into a concrete\ncost-optimization recommendation with a dollar-denominated savings\nestimate. The verdict is the highest-leverage action across six dimensions\n— concurrency tuning, MinSize\u002FMaxSize right-sizing, instance type sizing,\nhealth check optimization, VPC egress cost reduction, and pause\u002Fresume for\nnon-prod — applied in priority order. Always pairs the recommendation with\nexact CLI commands and a projected monthly savings figure.\n\n## Quick navigation\n\n| Section | What it covers | When to read it |\n|---|---|---|\n| **Quick reference** | Verdict thresholds (OPTIMIZED\u002FFURTHER_OPTIMIZATION_AVAILABLE) + optimization priority | Before any recommendation |\n| **Mindset** | Why concurrency is the primary lever, the MinSize baseline tradeoff, pause-for-savings | Understanding the cost model |\n| **Pre-flight** | Service metadata gate — auto-scaling config, instance config, traffic pattern | Before emitting any recommendation |\n| **Process** | Per-dimension analysis: concurrency, MinSize, instance type, health check, VPC, pause\u002Fresume | When choosing which optimization to apply |\n| **Output format** | Structured TARGET\u002FVERDICT\u002FREASON\u002FRECOMMENDATION\u002FSAVINGS template | Formatting the response |\n| **Anti-Patterns** | NEVER list — common mistakes that break scaling or increase cost | Review before risky changes |\n\n## Quick reference — verdict thresholds\n\n| Verdict | Trigger condition | Action |\n|---|---|---|\n| `OPTIMIZED` | All optimization dimensions checked; service configuration matches traffic pattern with no actionable savings > 5% | Emit confirmation, monitoring plan |\n| `FURTHER_OPTIMIZATION_AVAILABLE` | At least one dimension has actionable savings >= 5% or a structural change (pause\u002Fresume, instance type, concurrency) is recommended | Emit recommendation, CLI steps, estimated savings |\n\n**Optimization priority (apply in this order, each may yield independent\nsavings):**\n\n1. **Pause\u002Fresume for non-prod** — if the service is dev\u002Fstaging\u002FQA and\n   runs 24\u002F7, pausing outside business hours saves 100% of compute cost\n   (~70% of total App Runner bill). Highest single-action savings.\n2. **Concurrency tuning** — the primary lever. Higher concurrency = fewer\n   instances for the same traffic = lower cost. But higher concurrency\n   increases per-instance latency and memory pressure. Find the sweet spot.\n3. **MinSize right-sizing** — MinSize = always-provisioned baseline. For\n   static traffic, set MinSize to the steady-state instance count. For\n   dynamic traffic, set MinSize to 0 (scale to zero) or 1 (always-on).\n4. **Instance type sizing** — 1 vCPU\u002F2 GB is the cheapest; 2 vCPU\u002F4 GB\n   and 4 vCPU\u002F8 GB are 2x and 4x the cost. If CPU\u002Fmemory utilization is\n   low, the instance type may be overprovisioned. If concurrency is high\n   and latency is elevated, a larger instance type with higher concurrency\n   may be more cost-efficient.\n5. **Health check optimization** — longer intervals reduce false-positive\n   restarts but delay detection of unhealthy instances. Default 5s interval\n   is aggressive for most workloads.\n6. **VPC egress cost** — if the service uses a VPC connector, NAT gateway\n   charges apply to all egress. A VPC endpoint for AWS services (S3, DynamoDB,\n   etc.) eliminates NAT gateway cost for that traffic.\n\n**Cost baselines (2026, us-east-1):**\n\n| Instance type | Compute (per hour, provisioned) | Compute (per second, on-demand) | Best for |\n|---|---|---|---|\n| 1 vCPU \u002F 2 GB | $0.063\u002Fhr | $0.0000175\u002Fs | Low-traffic, small workloads |\n| 2 vCPU \u002F 4 GB | $0.126\u002Fhr | $0.0000350\u002Fs | Standard web APIs |\n| 4 vCPU \u002F 8 GB | $0.252\u002Fhr | $0.0000700\u002Fs | CPU\u002Fmemory-intensive workloads |\n\n- On-demand billing: per-second charges only when processing requests\n  (scale-to-zero eliminates idle cost).\n- Provisioned billing (MinSize >= 1): per-hour charges for the baseline\n  instances even when idle.\n- Concurrency does not have a direct charge — it controls how many\n  concurrent requests one instance handles, which indirectly controls\n  how many instances the autoscaler provisions.\n\n## Mindset\n\n**One-line takeaway:** Concurrency is the primary cost lever. Higher\nconcurrency means each instance handles more simultaneous requests, so\nthe autoscaler provisions fewer instances — but at the cost of increased\nper-request latency and memory pressure. Find the sweet spot where\nconcurrency is high enough to minimize instance count, but low enough to\nkeep latency within the SLO. Driven by three App Runner realities:\n\n- **Concurrency determines instance count.** App Runner scales based on\n  concurrent requests. With 100 concurrent requests and a concurrency\n  setting of 50, the autoscaler provisions 2 instances. With a concurrency\n  setting of 100, it provisions 1 instance. The cost difference is ~50%.\n  The latency difference depends on the workload — a CPU-bound request at\n  concurrency 100 takes longer than at concurrency 50 because the instance\n  is time-slicing.\n\n- **MinSize is the cost floor.** A MinSize of 1 means one instance is\n  always provisioned and billed at the hourly rate, even at 3am with zero\n  traffic. For dynamic traffic (web APIs with business-hours peaks), MinSize\n  0 (scale-to-zero) eliminates the idle cost. For static traffic (always-on\n  services), MinSize should match the steady-state instance count so the\n  autoscaler does not over-provision.\n\n- **Pause\u002Fresume for non-prod saves 100%.** App Runner services can be\n  paused (not deleted). A paused service incurs zero compute charges. For\n  dev\u002Fstaging\u002FQA environments that are only used during business hours,\n  pausing outside those hours saves ~75% of the monthly compute cost\n  (18 hours paused \u002F 24 hours). Combined with scale-to-zero during business\n  hours, non-prod services can run at a fraction of production cost.\n\n## Pre-flight: service metadata gate\n\nRun before classification. Misclassifying these produces wrong\nrecommendations.\n\n**Live-account pre-flight (skip if offline plan audit):**\nLive-account pre-flight command listing moved verbatim to [references\u002Fdiagnostic-commands.md](references\u002Fdiagnostic-commands.md).\nLoad on demand before a live-account optimization; offline plan audits skip it.\n\n**Malformed input:** if the input JSON is invalid or missing required\nfields, emit `VERDICT: ERROR` with `REASON: Service configuration is not\nvalid JSON or is missing required fields — cannot plan.` and\n`REMEDIATION: Re-fetch with aws apprunner describe-service --service-arn\n\u003Carn> --output json and re-plan.`\n\n| Service attribute | Effect on optimization |\n|---|---|\n| `Status: PAUSED` | No compute charges. Resume-related recommendation only. |\n| `MinSize: 0` (scale-to-zero) | Already optimized for idle cost. Check if cold-start latency is acceptable. |\n| `MinSize >= 1` | Baseline compute cost. Check if MinSize matches steady-state instance count. |\n| `Concurrency: 1` (minimum) | Maximum instance count = maximum cost. Almost always overprovisioned. |\n| `Concurrency: 1000` (maximum) | Minimum instance count = minimum cost. Check if latency is within SLO. |\n| `CPUUtilization \u003C 20%` (sustained) | Instance type may be overprovisioned. |\n| `CPUUtilization > 70%` (sustained) | Instance type may be underprovisioned, or concurrency too high. |\n| `MemoryUtilization > 80%` (sustained) | Concurrency too high or instance memory too low. Risk of OOM. |\n| `5xxResponseCount` spikes | Possible OOM from high concurrency or slow health checks. |\n| VPC connector with NAT gateway | Egress cost on all outbound traffic. |\n| `HealthCheckConfiguration.Interval: 1` | Aggressive health checking. May cause false-positive restarts. |\n| Custom domain with managed certificate | Negligible compute overhead. No optimization needed. |\n\n## Process — optimization analysis (apply in priority order)\n\n### Step 0: Expert knowledge — non-obvious App Runner autoscaling behaviors\n\nStep 0 expert-knowledge behaviors moved verbatim to [references\u002Fadvanced-patterns.md](references\u002Fadvanced-patterns.md).\nLoad on demand when a recommendation depends on non-obvious behavior (HPA stabilization window, non-linear concurrency\u002FCPU curve, resume cold-pull).\n\n### Step 1: Pause\u002Fresume analysis (non-prod only)\n\nIf the service is in a non-prod environment (dev\u002Fstaging\u002FQA):\n\n```\nIf service runs 24\u002F7:\n  Recommend pause outside business hours (e.g., 7pm-7am + weekends)\n  Savings = monthly compute * (paused_hours \u002F total_hours)\n  Typical: 18hr\u002Fday paused on weekdays + 48hr weekends = ~75% savings\n\nIf service is already paused or uses MinSize 0:\n  No pause\u002Fresume recommendation. Proceed to concurrency analysis.\n```\n\n**CLI for pause\u002Fresume automation:**\n\n```bash\n# Pause (using EventBridge scheduled rule)\naws apprunner pause-service --service-arn \u003Carn>\n\n# Resume\naws apprunner resume-service --service-arn \u003Carn>\n```\n\n### Step 2: Concurrency analysis\n\n```\ncurrent_concurrency = AutoScalingConfiguration.Concurrency\navg_concurrent_requests = max(RequestCount_per_5min \u002F 300) (peak concurrent)\navg_instance_count = average(InstanceCount over 30 days)\nobserved_concurrency_per_instance = avg_concurrent_requests \u002F avg_instance_count\n\nIf observed_concurrency_per_instance \u003C current_concurrency * 0.5:\n  → Concurrency is overprovisioned (instances are underutilized)\n  → Recommend: increase concurrency to reduce instance count\n  → New concurrency = current_concurrency * 1.5 (gradual increase)\n  → Savings = (current_instances - projected_instances) * hourly_rate * 730\n\nIf observed_concurrency_per_instance > current_concurrency * 0.8:\n  AND CPUUtilization > 60% OR MemoryUtilization > 75%:\n  → Concurrency is near the limit; increasing further risks latency\u002FOOM\n  → Recommend: increase instance type (more CPU\u002Fmemory per instance)\n  → Then increase concurrency to match\n\nIf latency p95 > SLO target:\n  → Concurrency may be too high for the workload\n  → Recommend: decrease concurrency to reduce per-request latency\n  → Cost increases but stays within SLO\n```\n\n### Step 3: MinSize\u002FMaxSize right-sizing\n\n```\nIf MinSize > 0:\n  steady_state_instances = percentile(InstanceCount, 10) (low-traffic baseline)\n  If MinSize > steady_state_instances:\n    → MinSize is overprovisioned; reduce to steady_state_instances\n    → If steady_state_instances = 0: recommend MinSize 0 (scale-to-zero)\n\nIf MaxSize \u003C peak(InstanceCount):\n  → MaxSize is too low; instances are hitting the cap during peaks\n  → Recommend: increase MaxSize to peak(InstanceCount) * 1.5\n\nIf MaxSize > peak(InstanceCount) * 3:\n  → MaxSize is excessively high; no immediate cost impact but risk of\n    runaway scaling\n  → Recommend: reduce MaxSize to peak(InstanceCount) * 2\n```\n\n### Step 4: Instance type analysis\n\n```\ncpu_avg = average(CPUUtilization over 30 days)\nmem_avg = average(MemoryUtilization over 30 days)\n\nIf cpu_avg \u003C 25% AND mem_avg \u003C 40%:\n  AND current instance type > 1 vCPU \u002F 2 GB:\n  → Instance type is overprovisioned\n  → Recommend: downgrade to 1 vCPU \u002F 2 GB\n  → Savings = (current_hourly - new_hourly) * instance_count * 730\n\nIf cpu_avg > 70% OR mem_avg > 80%:\n  → Instance type is underprovisioned\n  → Recommend: upgrade to next size\n  → Also consider: increasing concurrency with larger instance type\n```\n\n### Step 5: Health check optimization\n\n```\nIf HealthCheckConfiguration.Interval \u003C 5:\n  AND InstanceRestartCount is high:\n  → Health check interval is too aggressive; false-positive restarts\n  → Recommend: increase interval to 10-15s\n\nIf HealthCheckConfiguration.Timeout > Interval * 0.5:\n  → Timeout is too close to interval; health checks may always timeout\n  → Recommend: set timeout to Interval * 0.3\n\nIf HealthCheckConfiguration.UnhealthyThreshold \u003C 3:\n  → Single failed check triggers replacement; too sensitive\n  → Recommend: set to 3-5\n```\n\n### Step 6: VPC egress cost analysis\n\n```\nIf NetworkConfiguration.EgressConfiguration uses NAT gateway:\n  nat_cost = CostExplorer filter \"NatGateway\" for the VPC\n  If nat_cost > compute_cost * 0.3:\n    → NAT gateway cost is significant\n    → Recommend: add VPC endpoints for AWS service traffic\n    → Savings = nat_cost * (aws_service_traffic_fraction)\n```\n\n### Step 7: Synthesize — emit verdict\n\nIf any dimension yielded actionable savings >= 5%:\n→ `VERDICT: FURTHER_OPTIMIZATION_AVAILABLE`\n\nIf all dimensions are optimal (no actionable savings >= 5%):\n→ `VERDICT: OPTIMIZED`\n\n## Output format (per service)\n\n```text\nTARGET: \u003Cservice-name> (region: \u003Cregion>)\nVERDICT: OPTIMIZED | FURTHER_OPTIMIZATION_AVAILABLE\nREASON: \u003Cone-line summary of the primary finding>\nRECOMMENDATION:\n  - \u003Cspecific recommendation 1>\n  - \u003Cspecific recommendation 2>\nESTIMATED_SAVINGS: $\u003Cmonthly_savings>\u002Fmonth (\u003Cpercentage>% of current spend)\nMIGRATION_STEPS:\n  1. \u003CCLI command or action>\n  2. \u003CCLI command or action>\n  3. \u003Cverification step>\nCURRENT_MONTHLY_COST: $\u003Camount>\nPROJECTED_MONTHLY_COST: $\u003Camount>\nTRAFFIC_PATTERN: \u003Cstatic | dynamic | bursty | intermittent>\nRISK: \u003CLOW | MODERATE | HIGH>\nNOTES: \u003Ccaveats, monitoring recommendations>\n```\n\n### Worked example — concurrency tuning (FURTHER_OPTIMIZATION_AVAILABLE)\n\n```text\nTARGET: prod-api-service (region: us-east-1)\nVERDICT: FURTHER_OPTIMIZATION_AVAILABLE\nREASON: Concurrency 100 is overprovisioned — observed peak is 42 concurrent\n        requests per instance, meaning 58% of provisioned capacity is unused.\nRECOMMENDATION:\n  - Increase concurrency from 100 to 150 (gradual 50% increase)\n  - Current avg 2.3 instances → projected 1.5 instances at concurrency 150\n  - Monitor latency p95 for 7 days post-change\nESTIMATED_SAVINGS: $230\u002Fmonth (19% of current compute spend)\nMIGRATION_STEPS:\n  1. Create new auto-scaling configuration:\n     aws apprunner create-auto-scaling-configuration \\\n       --auto-scaling-configuration-name prod-api-service-asg-v2 \\\n       --min-size 1 \\\n       --max-size 10 \\\n       --concurrency 150\n  2. Apply to the service:\n     aws apprunner update-service \\\n       --service-arn arn:aws:apprunner:us-east-1:111111111111:service\u002Fprod-api-service\u002Fabc \\\n       --auto-scaling-configuration-arn arn:aws:apprunner:us-east-1:111111111111:autoscalingconfiguration\u002Fprod-api-service-asg-v2\u002F1\n  3. Monitor for 7 days:\n     aws cloudwatch get-metric-statistics \\\n       --namespace AWS\u002FAppRunner \\\n       --metric-name Latency \\\n       --dimensions Name=ServiceId,Value=\u003Cid> \\\n       --statistics Average \\\n       --period 300 --start-time ... --end-time ...\n  4. If latency p95 increases > 20%, revert to concurrency 100\nCURRENT_MONTHLY_COST: $1,200\u002Fmonth\nPROJECTED_MONTHLY_COST: $970\u002Fmonth\nTRAFFIC_PATTERN: dynamic (3x peak during business hours)\nRISK: MODERATE\nNOTES:\n  - Concurrency 150 is still below the workload's CPU limit (avg\n    CPUUtilization at concurrency 100 is 35%; at 150 it is projected\n    ~52%).\n  - If latency degrades, consider upgrading to 2 vCPU \u002F 4 GB instead\n    of reducing concurrency back.\n```\n\n### Worked example — pause\u002Fresume for non-prod (FURTHER_OPTIMIZATION_AVAILABLE)\n\nSecondary worked example moved verbatim to [references\u002Fworked-examples.md](references\u002Fworked-examples.md).\nLoad on demand when emitting a pause\u002Fresume recommendation.\n\n### Worked example — already optimal (OPTIMIZED)\n\nSecondary worked example moved verbatim to [references\u002Fworked-examples.md](references\u002Fworked-examples.md).\nLoad on demand when confirming a service is already optimal.\n\n## Anti-Patterns — NEVER\n\n- NEVER increase concurrency without monitoring latency for at least 7 days\n  post-change. Higher concurrency reduces instance count but increases per-\n  request latency on CPU-bound workloads. A blind increase from 50 to 200\n  can degrade p95 latency by 3-5x.\n\n- NEVER set MinSize to 0 for a production service that requires low cold-\n  start latency. App Runner cold start after scale-to-zero is 5-30 seconds\n  (container pull + start). For user-facing APIs, keep MinSize at 1.\n\n- NEVER pause a production service. Pause is for non-prod only. Pausing\n  removes the endpoint and takes the service offline.\n\n- NEVER downgrade instance type based on a single day of low utilization.\n  CPU\u002Fmemory utilization varies by day of week and time of day. Use at\n  least 14 days of data, ideally 30, to capture the full traffic cycle.\n\n- NEVER set MaxSize below the peak observed instance count. The autoscaler\n  will hit the cap and return 5xx errors during traffic spikes.\n\n- NEVER ignore VPC egress cost. For services with a VPC connector, NAT\n  gateway charges ($0.045\u002FGB processed + $0.045\u002FGB data transfer) can\n  exceed compute cost for high-egress workloads. Use VPC endpoints for\n  AWS service traffic.\n\n- NEVER assume App Runner is the right platform for all workloads. For\n  sustained high-traffic services (> $2,000\u002Fmonth), ECS\u002FFargate or EC2 may\n  be more cost-efficient. Run a cost-per-request comparison.\n\n- NEVER change concurrency, MinSize, and instance type simultaneously. Make\n  one change at a time and measure the impact for 7 days before the next.\n  Simultaneous changes make it impossible to attribute cost\u002Flatency\n  changes to a specific dimension.\n\n- NEVER delete an auto-scaling configuration that is still referenced by a\n  service. `delete-auto-scaling-configuration` fails if the configuration\n  is in use. Use `update-service` to switch to the new configuration first.\n\n- NEVER set health check interval to 1 second unless the workload requires\n  instant failure detection. A 1-second interval with a 3-second timeout\n  and 3 unhealthy threshold means 3 seconds to detect a failure — but it\n  also means transient latency spikes (a slow request holding the health\n  check thread) trigger false-positive restarts. Use 5-10 seconds.\n\n- NEVER reduce concurrency to reduce latency without checking if the\n  workload is CPU-bound or I\u002FO-bound. For I\u002FO-bound workloads (API calls,\n  database queries), concurrency has minimal latency impact — the instance\n  handles concurrent I\u002FO natively. For CPU-bound workloads, concurrency\n  directly increases latency.\n\n- NEVER assume scale-to-zero eliminates all cost. App Runner charges for\n  container image storage in ECR and any custom domain certificate\n  management. These are typically negligible (\u003C $5\u002Fmonth) but not zero.\n\n- NEVER forget that deployments pause autoscaling. During a deployment, the\n  autoscaler does not add or remove instances. High-traffic deployments may\n  cause temporary latency spikes. Schedule deployments during low-traffic\n  windows.\n\n- NEVER auto-execute a state-changing App Runner CLI without the CONFIRM\n  gate. Changing the auto-scaling configuration affects production traffic.\n  The service briefly transitions during configuration changes.\n\n## Pre-flight safety checks (run before any optimization CLI)\n\nPre-flight safety checks (CONFIRMATION GATE, pre-state capture, post-change monitoring) moved verbatim to [references\u002Fdiagnostic-commands.md](references\u002Fdiagnostic-commands.md).\nLoad on demand before executing any state-changing optimization CLI.\n\n## Recent AWS features (2024-2026)\n\nRecent AWS features (2024-2026) moved verbatim to [references\u002Fadvanced-patterns.md](references\u002Fadvanced-patterns.md).\nLoad on demand when evaluating v2 native VPC networking, blue\u002Fgreen deployments, or the observability dashboard.\n\n## References (load on demand)\n\n- [references\u002Fworked-examples.md](references\u002Fworked-examples.md) — secondary worked examples moved from SKILL.md: pause\u002Fresume for non-prod, already-optimal OPTIMIZED output.\n- [references\u002Fadvanced-patterns.md](references\u002Fadvanced-patterns.md) — Step 0 non-obvious autoscaling behaviors and Recent AWS features (2024-2026) moved from SKILL.md.\n- [references\u002Fdiagnostic-commands.md](references\u002Fdiagnostic-commands.md) — live-account pre-flight command listing and pre-flight safety checks moved from SKILL.md.\n- [references\u002Fapprunner-pricing-and-sizing.md](references\u002Fapprunner-pricing-and-sizing.md) — pricing model, instance-type pricing, monthly cost examples, sizing heuristics.\n- [references\u002Foptimization-procedures.md](references\u002Foptimization-procedures.md) — per-dimension optimization procedures (concurrency, MinSize, instance type, health check, VPC, pause\u002Fresume).\n\n## Domain\n\nAWS CloudOps \u002F App Runner Autoscaling Cost Optimization & FinOps.\n\n## AWS documentation\n\n- **AWS App Runner Developer Guide** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fwhat-is-apprunner.html\n- **App Runner auto-scaling** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-autoscaling.html\n- **App Runner autoscaling configuration** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-autoscaling.html#manage-autoscaling.config\n- **App Runner instance configuration** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-configure.html\n- **App Runner VPC networking** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fnetwork-vpc.html\n- **App Runner health checks** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-configure.html#manage-configure.health\n- **App Runner API Reference** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fapi\u002FWelcome.html\n- **App Runner pricing** — https:\u002F\u002Faws.amazon.com\u002Fapprunner\u002Fpricing\u002F\n- **App Runner pause\u002Fresume** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmanage-delete.html#manage-delete.pause\n- **CloudWatch App Runner metrics** — https:\u002F\u002Fdocs.aws.amazon.com\u002Fapprunner\u002Flatest\u002Fdg\u002Fmonitor-cw.html\n",{"name":5,"license":15,"metadata":24,"description":7,"compatibility":44},{"tags":25,"phase":26,"author":27,"domain":28,"family":29,"version":30,"keywords":31,"task_type":32,"complexity":33,"entry_point":34,"skill_class":35,"when_to_use":36,"requires_llm":37,"verdict_shape":38,"when_not_to_use":39,"lifecycle_status":40,"invocation_schema":41,"supports_pipeline":37,"invocation_example":42,"activation_triggers":43},"aws, apprunner, compute, cost-optimization, autoscaling, finops, concurrency, serverless-container, optimize","3","Jacky Chan — AWS Community Builder","aws-cloudops","Compute","0.1.0","App Runner, auto-scaling configuration, concurrency, MinSize, MaxSize, provisioned instances, instance type, 1 vCPU 2 GB, 2 vCPU 4 GB, 4 vCPU 8 GB, CPU utilization, memory utilization, health check, health check interval, scale-in cooldown, scale-out speed, static traffic, dynamic traffic, cost per request, custom domain, SSL certificate, deployment, VPC ingress, VPC egress, NAT gateway, observability cost, CloudWatch Logs, pause service, resume service, non-prod savings, FinOps","optimize","medium","false","capability","Optimising App Runner autoscaling configuration, tuning concurrency settings, right-sizing MinSize\u002FMaxSize, evaluating instance type upgrades (1\u002F2\u002F4 vCPU), analyzing cost-per-request, deciding whether to pause non-prod services, reviewing VPC egress cost, tuning health check intervals, or conducting a FinOps review of App Runner services.","true","OPTIMIZED | FURTHER_OPTIMIZATION_AVAILABLE","ECS\u002FFargate service autoscaling (use an ECS autoscaling skill), Lambda cost optimization (use lambda-cost-optimizer), EC2 instance rightsizing (use ec2-rightsizing-optimizer), or App Runner troubleshooting (deployment failures, configuration errors — use the App Runner troubleshooter). This skill focuses on cost-driven autoscaling optimization, not functional debugging.","active","Input: either (a) an App Runner service ARN + live-account context, (b) a service configuration document (describe-service + describe-auto-scaling-configuration output), OR (c) CloudWatch metrics (RequestCount, InstanceCount, CPUUtilization, MemoryUtilization, 4xx, 5xx) with at least 14 days of observation. Output: a deterministic TARGET\u002FVERDICT\u002FREASON\u002FRECOMMENDATION\u002FESTIMATED_SAVINGS\u002FMIGRATION_STEPS block per service, where VERDICT is one of OPTIMIZED, FURTHER_OPTIMIZATION_AVAILABLE.","# Minimal valid input (offline finding classification): ServiceName: prod-api-service AutoScalingConfiguration:\n  MinSize: 1\n  MaxSize: 25\n  Concurrency: 100\nInstanceConfiguration:\n  Cpu: 1 vCPU\n  Memory: 2 GB\nRegion: us-east-1 Metrics (last 30 days):\n  - RequestCount: 8,000,000\u002Fmonth\n  - Avg InstanceCount: 2.3 (MinSize 1 is sufficient)\n  - Avg CPUUtilization: 15%\n  - Avg MemoryUtilization: 22%\n  - 5xx rate: 0.01%\nTraffic pattern: dynamic (peak 3x baseline during business hours) Cost (last 30 days): $1,200\u002Fmonth Emit the standard optimization block (TARGET, VERDICT, REASON, RECOMMENDATION, ESTIMATED_SAVINGS, MIGRATION_STEPS).","optimise App Runner, App Runner autoscaling, App Runner concurrency, App Runner MinSize, App Runner MaxSize, App Runner instance type, App Runner cost, App Runner FinOps, App Runner pause, App Runner resume, App Runner VPC egress, App Runner health check, App Runner scale-in cooldown, App Runner cost per request, App Runner right-size, App Runner provisioned, reduce App Runner bill, App Runner non-prod savings","Agent runtime that reads SKILL.md (Claude Code, Cursor, Windsurf, Codex, Gemini). Offline recommendation classification works from pasted App Runner service configuration and CloudWatch metrics. Live-account optimization uses aws apprunner describe-service, describe-auto-scaling- configuration, list-services, list-auto-scaling-configurations, aws cloudwatch get-metric-statistics (RequestCount, InstanceCount, CPUUtilization, MemoryUtilization, 4xx, 5xx, Latency), aws ce get-cost- and-usage...","main","eeace51bd5dd1db2030729d83eeab988861553aa","skills\u002Fapprunner-autoscaling-optimizer",0,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills","https:\u002F\u002Fgithub.com\u002Fjhkchan","CTO & Co-Founder @Beever-AI  & @votee. CAISP. AWS Community Builder. Organizer@GDGCloudHK. MongoDB Champion. Technopreneur + Infornographer.","2026-09-24 08:54:10",{"source":8,"skillName":5,"commands":54,"prompt":64,"zipAvailable":65},[55,58,61],{"tool":56,"command":57},"npx","npx skills add jhkchan\u002Faws-agent-skills --skill apprunner-autoscaling-optimizer",{"tool":59,"command":60},"bunx","bunx skills add jhkchan\u002Faws-agent-skills --skill apprunner-autoscaling-optimizer",{"tool":62,"command":63},"pnpm","pnpm dlx skills add jhkchan\u002Faws-agent-skills --skill apprunner-autoscaling-optimizer","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002FSKILL.md",true,[67,73,77,81,85,89,93,97,101,105,109,113,117,121,125,129,133,137,141,145],{"path":68,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":71,"sourceHtmlUrl":72},"eval\u002Ftest-cases.yaml","REFERENCE","参考文档",7073,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Feval\u002Ftest-cases.yaml",{"path":74,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":75,"sourceHtmlUrl":76},"evals\u002Fbaselines\u002Falready-optimal-config-optimized.md",552,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fbaselines\u002Falready-optimal-config-optimized.md",{"path":78,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":79,"sourceHtmlUrl":80},"evals\u002Fbaselines\u002Fconcurrency-tuning-further-optimization.md",818,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fbaselines\u002Fconcurrency-tuning-further-optimization.md",{"path":82,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":83,"sourceHtmlUrl":84},"evals\u002Fbaselines\u002Finstance-type-downgrade-further-optimization.md",637,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fbaselines\u002Finstance-type-downgrade-further-optimization.md",{"path":86,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":87,"sourceHtmlUrl":88},"evals\u002Fbaselines\u002Fpause-non-prod-savings-further-optimization.md",565,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fbaselines\u002Fpause-non-prod-savings-further-optimization.md",{"path":90,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":91,"sourceHtmlUrl":92},"evals\u002Fbaselines\u002Fvpc-egress-nat-cost-further-optimization.md",853,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fbaselines\u002Fvpc-egress-nat-cost-further-optimization.md",{"path":94,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":95,"sourceHtmlUrl":96},"evals\u002Feval_config.yaml",2068,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Feval_config.yaml",{"path":98,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":99,"sourceHtmlUrl":100},"evals\u002Fevals.json",4485,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fevals.json",{"path":102,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":103,"sourceHtmlUrl":104},"evals\u002Fprompts\u002Falready-optimal-config-optimized.md",1113,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fprompts\u002Falready-optimal-config-optimized.md",{"path":106,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":107,"sourceHtmlUrl":108},"evals\u002Fprompts\u002Fconcurrency-tuning-further-optimization.md",973,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fprompts\u002Fconcurrency-tuning-further-optimization.md",{"path":110,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":111,"sourceHtmlUrl":112},"evals\u002Fprompts\u002Finstance-type-downgrade-further-optimization.md",921,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fprompts\u002Finstance-type-downgrade-further-optimization.md",{"path":114,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":115,"sourceHtmlUrl":116},"evals\u002Fprompts\u002Fpause-non-prod-savings-further-optimization.md",793,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fprompts\u002Fpause-non-prod-savings-further-optimization.md",{"path":118,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":119,"sourceHtmlUrl":120},"evals\u002Fprompts\u002Fvpc-egress-nat-cost-further-optimization.md",1184,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fevals\u002Fprompts\u002Fvpc-egress-nat-cost-further-optimization.md",{"path":122,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":123,"sourceHtmlUrl":124},"examples\u002FREADME.md",6477,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fexamples\u002FREADME.md",{"path":126,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":127,"sourceHtmlUrl":128},"references\u002Fadvanced-patterns.md",10498,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Freferences\u002Fadvanced-patterns.md",{"path":130,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":131,"sourceHtmlUrl":132},"references\u002Fapprunner-pricing-and-sizing.md",7425,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Freferences\u002Fapprunner-pricing-and-sizing.md",{"path":134,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":135,"sourceHtmlUrl":136},"references\u002Fdiagnostic-commands.md",3071,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Freferences\u002Fdiagnostic-commands.md",{"path":138,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":139,"sourceHtmlUrl":140},"references\u002Foptimization-procedures.md",11436,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Freferences\u002Foptimization-procedures.md",{"path":142,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":143,"sourceHtmlUrl":144},"references\u002Fworked-examples.md",2854,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Freferences\u002Fworked-examples.md",{"path":146,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":147,"sourceHtmlUrl":148},"skill.usf.yaml",4659,"https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fapprunner-autoscaling-optimizer\u002Fskill.usf.yaml",[150,157,164,171,178,185],{"skillNo":151,"name":152,"nameZh":6,"description":153,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":154,"categories":155,"keywords":156,"publishedAt":19},"01M7ADVD60X4PCAPKSE58P9NDC","xray-tracing-deployer","Deploys AWS X-Ray distributed tracing with production-grade configuration: X-Ray daemon deployment (EC2 systemd, ECS sidecar, EKS DaemonSet, Lambda built-in), SDK instrumentation (Java, Python, Node.js, Go, .NET) with annotations and metadata, sampling rules (default reservoir + rate vs custom rules with URL \u002F method \u002F service-name predicates), service map generation, trace query and retrieval, groups and insights, CloudWatch ServiceLens integration, IAM permissions (PutTraceSegments, GetSamplingRules), and latest features (Lambda Powertools tracing, OpenTelemetry \u002F ADOT Collector support, W3C trace context). Emits a READY_TO_DEPLOY checklist with every configuration item verified. Use when enabling X-Ray on a new application, adding tracing to ECS \u002F EKS \u002F Lambda, validating a sampling configuration, or generating daemon deployment CLI and IaC templates. Triggers: X-Ray, distributed tracing, X-Ray daemon, sampling rules, service map, X-Ray SDK, OpenTelemetry, ADOT, Lambda Powertools, CloudWatch ServiceLens.","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fxray-tracing-deployer\u002FSKILL.md",[],[],{"skillNo":158,"name":159,"nameZh":6,"description":160,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":161,"categories":162,"keywords":163,"publishedAt":19},"01M7ADTVJ7DCBM9CB1FY51GH7J","wellarchitected-workload-auditor","Audits AWS Well-Architected Tool workloads for review staleness, per-pillar high-risk issue counts, milestone tracking gaps, and remediation plan completeness. Emits a deterministic verdict (STALE_REVIEW | HIGH_RISK | CONFIG_GAP | OK) per workload with per-pillar risk breakdown, milestone trend analysis, and specific remediation. Use when reviewing Well-Architected workloads, checking WA review freshness, auditing high-risk issues per pillar, validating milestone coverage, or tracking remediation plans.","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwellarchitected-workload-auditor\u002FSKILL.md",[],[],{"skillNo":165,"name":166,"nameZh":6,"description":167,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":168,"categories":169,"keywords":170,"publishedAt":19},"01M7ADT59G6FTTNZYHT2QRPNNP","wellarchitected-review-operator","Operates AWS Well-Architected Tool reviews end-to-end — creates workloads, runs pillar reviews (operational excellence, security, reliability, performance, cost optimization, sustainability, Prosperity), answers questions with risk-tier rationale, generates reports and improvement plans, creates milestones, and integrates with Trusted Advisor and Well-Architected Labs. Runs deterministic pre-checks (workload existence, lens availability, pillar coverage), emits the exact update-answer \u002F get-consolidated-report CLI behind a CONFIRM gate, and verifies risk-tier transitions. Emits READY | BLOCKED | COMPLETED. Use when creating a workload, answering pillar questions, generating an improvement plan, creating a milestone, or integrating with Trusted Advisor. Triggers: well-architected tool, well architected review, workload review, pillar, operational excellence, security, reliability, performance, cost optimization, sustainability, prosperity, improvement plan, milestone, trusted advisor.","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwellarchitected-review-operator\u002FSKILL.md",[],[],{"skillNo":172,"name":173,"nameZh":6,"description":174,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":175,"categories":176,"keywords":177,"publishedAt":19},"01M7ADSH5JA9QHNPPHQCD7EVA7","wafv2-web-acl-deployer","Provisions AWS WAFv2 Web ACLs with correct production defaults: scope selection (CLOUDFRONT vs REGIONAL), managed rule groups (AWSManagedRulesCommonRuleSet, SQLiRuleSet, LinuxRuleSet, WindowsRuleSet, IPReputationList, BotControlRuleSet, AmazonIpReputationList, ATP), custom rules (byte match, IP set, regex pattern, geo match, size constraint, rate-based), rule priority ordering, logging destinations (Kinesis Firehose, CloudWatch Logs, S3), resource association (ALB, API Gateway, CloudFront), rate-based rules with aggregate keys (IP, forwarded IP), Challenge and CAPTCHA actions, and account takeover prevention. Emits a READY_TO_DEPLOY checklist. Use when creating a Web ACL, attaching managed rules, configuring rate limiting, enabling WAF logging, associating a Web ACL with an ALB\u002FAPI Gateway\u002FCloudFront, or configuring CAPTCHA\u002FChallenge actions. Triggers: create WAF Web ACL, WAFv2, managed rule groups, WAF rate limiting, WAF logging, WAF CAPTCHA, WAF Challenge action, bot control, ATP managed rules.","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwafv2-web-acl-deployer\u002FSKILL.md",[],[],{"skillNo":179,"name":180,"nameZh":6,"description":181,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":182,"categories":183,"keywords":184,"publishedAt":19},"01M7ADRYFAT8R73FBNCHRC573X","wafv2-web-acl-auditor","Audits AWS WAFv2 Web ACL configurations to determine whether the ACL provides effective protection: default-action posture (Allow vs Block), managed-rule-group coverage gaps, rule effectiveness (BLOCK vs COUNT, shadow\u002Fbypass rules, stale exclusions), rate-based rule correctness, logging and visibility configuration, and text-transformation bypass vectors. Emits a deterministic verdict per Web ACL. Use when reviewing a WAFv2 Web ACL for security posture, checking managed-rule coverage, validating rate-limit configuration, auditing WAF logging, or hardening a WAF before production deployment. Triggers: WAF, WAFv2, Web ACL, managed rule group, AWSManagedRules, rate-based rule, CAPTCHA, Challenge, OverrideAction, Count mode, forwarded IP, Bot Control, ATP, text transformation, WAF logging, WAF visibility, OWASP, SQLi, XSS, RFI.","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwafv2-web-acl-auditor\u002FSKILL.md",[],[],{"skillNo":186,"name":187,"nameZh":6,"description":188,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":189,"categories":190,"keywords":191,"publishedAt":19},"01M7ADR8W7G72JX4KFGN03QK8Q","waf-rule-deployer","Provisions AWS WAFv2 rule sets with production defaults: Web ACL creation (CloudFront scope in us-east-1 vs Regional scope), rule groups (managed vs custom) with capacity planning in WCU (Web ACL Capacity Units, 1500 default per ACL), managed rule groups (AWSManagedRulesCommonRuleSet, AWSManagedRulesSQLiRuleSet, AWSManagedRulesAmazonIpReputationList, AWSManagedRulesLinuxRuleSet, AWSManagedRulesWindowsRuleSet, AWSManagedRulesUnixRuleSet), custom rules (byte-match, regex pattern set, geo-match, rate-based), IP set and regex pattern set resources, rule priority ordering (lowest number evaluated first), action types (allow, block, count, CAPTCHA, Challenge), bot control managed rule group, account takeover. Triggers: create WAF rule, WAF managed rule group, WAF custom rule, WAF byte match, WAF geo match, WAF rate based rule, WAF IP set, WAF regex pattern set, WAF WCU budget, WAF CloudFront ACL, WAF logging Kinesis Firehose, WAF bot control, WAF ATP, WAF label matching, WAF Challenge action, WAF CAPTCHA.","https:\u002F\u002Fgithub.com\u002Fjhkchan\u002Faws-agent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwaf-rule-deployer\u002FSKILL.md",[],[]]