AGENT SKILL

rds-failover-troubleshooter

Diagnoses RDS and Aurora failover issues through a thirteen-category diagnostic tree: Multi-AZ failover not triggering (health check threshold too conservative), failover taking too long (DNS propagation, recovery mode full), Aurora writer endpoint not redirecting (application using instance endpoint instead of cluster endpoint), read replica promotion failures (network connectivity, storage), storage-full preventing failover (allocated storage exhausted), parameter group mismatch between primary and standby, option group conflicts preventing standby from taking over, Aurora cluster endpoint type confusion (writer vs reader vs custom vs instance), failover priority tier misconfiguration (tier 0 vs tier 1 not set on the intended promotion target), Aurora Global DB failover (managed vs unmanaged), RDS Proxy failover behavior (connection pool not draining), application connection string not using cluster endpoint, and CloudWatch failover event gaps. Walks symptoms to a verified root cause with...

License 许可
Apache-2.0
最近更新
2026-09-24
Stars
★ 4
Forks
0

01 前言 Front Matter

name
rds-failover-troubleshooter
license
Apache-2.0
metadata
{ "tags": "rds, aurora, databases, troubleshooting, failover, multi-az, high-availability, cluster-endpoint, rds-proxy", "phase": "2", "author": "Jacky Chan — AWS Community Builder", "domain": "aws-cloudops", "family": "Databases", "version": "0.1.0", "keywords": "RDS, Aurora, failover, Multi-AZ, writer endpoint, cluster endpoint, read replica promotion, storage-full, parameter group mismatch, option group conflict, failover priority, tier 0, Aurora Global DB, RDS Proxy, connection pool, DNS propagation, CloudWatch failover events, troubleshooting", "task_type": "troubleshoot", "complexity": "high", "entry_point": "false", "skill_class": "capability", "when_to_use": "Diagnosing an RDS or Aurora failover issue (Multi-AZ failover not triggering, failover taking too long, writer endpoint not redirecting, read replica promotion failure, storage-full preventing failover, parameter group mismatch, option group conflict, failover priority misconfiguration, Aurora Global DB failover, RDS Proxy failover, application connection issues), walking a symptom to the failed layer with verify commands, or triaging a \"failover did not work\" incident where the root cause may be endpoint configuration, health threshold, storage, parameter group, option group, failover tier, or application connection string — not necessarily the database engine itself.", "requires_llm": "true", "verdict_shape": "ROOT_CAUSE_IDENTIFIED | INSUFFICIENT_DATA", "when_not_to_use": "RDS connectivity debugging (use rds-connectivity-troubleshooter), RDS performance tuning (use rds-cost-optimizer for right-sizing), Aurora backup and restore (use rds-backup-restore-operator), or RDS instance creation (use rds-instance-deployer). This skill diagnoses failover-time issues; it does not tune query performance or audit steady-state security posture.", "lifecycle_status": "active", "invocation_schema": "Input: either (a) a symptom description (error message, observed behaviour, \"failover took 5 minutes\", \"writer endpoint still points to old instance\"), optionally paired with the cluster configuration and recent RDS events, OR (b) a DB cluster identifier plus caller context (application connection string, observed error) for live-account diagnosis. Output: a deterministic TARGET/VERDICT/REASON/ROOT_CAUSE/EVIDENCE/REMEDIATION block where VERDICT ∈ {ROOT_CAUSE_IDENTIFIED, INSUFFICIENT_DATA} and ROOT_CAUSE ∈ {APP_NOT_CLUSTER_ENDPOINT, MULTI_AZ_HEALTH_THRESHOLD, FAILOVER_PRIORITY_TIER, STORAGE_FULL, PARAMETER_GROUP_MISMATCH, OPTION_GROUP_CONFLICT, DNS_PROPAGATION_DELAY, CONNECTION_POOL_CACHING, AURORA_GLOBAL_FAILOVER_MODE, RDS_PROXY_POOL_DRAIN, RECOVERY_MODE_FULL, CLOUDWATCH_EVENT_GAP, READ_REPLICA_PROMOTION_FAILURE, UNKNOWN}.", "supports_pipeline": "true", "invocation_example": "# Minimal valid input (offline symptom classification):\nSymptom: \"Aurora PostgreSQL cluster prod-db-cluster\nfailed over automatically at 03:17 UTC. The writer\nendpoint resolved to the new writer within 1 second,\nbut the application continued hitting the old writer\nfor 90 seconds, then connection errors cleared.\"\nDBClusterIdentifier: prod-db-cluster\nEngine: aurora-postgresql\nWriterEndpoint: prod-db-cluster.cluster-xxxxxxxxxxxx.us-east-1.rds.amazonaws.com\nApplication connection string: uses the instance endpoint\nprod-db-cluster-instance-1.xxxxxxxxxxxx.us-east-1.rds.amazonaws.com\ninstead of the cluster endpoint", "activation_triggers": "RDS failover not working, Aurora failover not triggering, Multi-AZ failover failed, Aurora writer endpoint not redirecting, RDS failover taking too long, Aurora failover priority, read replica promotion failed, storage-full failover, parameter group mismatch failover, option group conflict failover, Aurora Global DB failover, RDS Proxy failover, application not connecting after failover, Aurora cluster endpoint confusion, troubleshoot RDS failover" }
description
Diagnoses RDS and Aurora failover issues through a thirteen-category diagnostic tree: Multi-AZ failover not triggering (health check threshold too conservative), failover taking too long (DNS propagation, recovery mode full), Aurora writer endpoint not redirecting (application using instance endpoint instead of cluster endpoint), read replica promotion failures (network connectivity, storage), storage-full preventing failover (allocated storage exhausted), parameter group mismatch between primary and standby, option group conflicts preventing standby from taking over, Aurora cluster endpoint type confusion (writer vs reader vs custom vs instance), failover priority tier misconfiguration (tier 0 vs tier 1 not set on the intended promotion target), Aurora Global DB failover (managed vs unmanaged), RDS Proxy failover behavior (connection pool not draining), application connection string not using cluster endpoint, and CloudWatch failover event gaps. Walks symptoms to a verified root cause with...
compatibility
Agent runtime that reads SKILL.md (Claude Code, Cursor, Windsurf, Codex, Gemini). Offline symptom classification works from pasted error messages and cluster configuration. Live-account diagnosis uses aws rds describe-db-clusters, aws rds describe-db-instances, aws rds describe-db-cluster-endpoints, aws rds describe-db-cluster-parameter-groups, aws rds describe-events, aws cloudwatch get-metric-statistics, aws ec2 describe-route-tables, aws ec2 describe-security-groups, and aws logs...

02 资源文件 21

03 SKILL.md 正文

RDS Failover Troubleshooter

Quick start

Moved verbatim to references/advanced-patterns.md — load on demand.

Mindset

Moved verbatim to references/advanced-patterns.md — load on demand.

Philosophy

Moved verbatim to references/advanced-patterns.md — load on demand.

Quick reference — symptom triage table

Symptom phrase / error Most likely root cause First probe
Failover triggered but app still hits old instance APP_NOT_CLUSTER_ENDPOINT Check application connection string for instance vs cluster endpoint
App recovers 2-5 minutes after failover despite DNS TTL 1s CONNECTION_POOL_CACHING Check DB driver connection pool TTL, JVM DNS cache (networkaddress.cache.ttl)
Failover did not trigger at all MULTI_AZ_HEALTH_THRESHOLD describe-db-instances for Multi-AZ config; CloudWatch DatabaseConnection and DiskQueueDepth metrics
New writer is not the expected instance FAILOVER_PRIORITY_TIER describe-db-instances for PromotionTier on each instance
InsufficientStorage or FailoverFailed STORAGE_FULL describe-db-instances for StorageAllocated vs StorageUsed; CloudWatch FreeStorageSpace
Standby cannot take over; error in RDS events PARAMETER_GROUP_MISMATCH or OPTION_GROUP_CONFLICT describe-db-instances for DBParameterGroups and OptionGroupMemberships on primary vs standby
Aurora writer endpoint not resolving to new writer DNS_PROPAGATION_DELAY nslookup or dig on the cluster endpoint; check DNS resolver cache
Read replica promotion fails READ_REPLICA_PROMOTION_FAILURE describe-db-instances for replica status; check network connectivity and storage
Aurora Global DB failover not working AURORA_GLOBAL_FAILOVER_MODE describe-global-clusters; check if managed or unmanaged failover
RDS Proxy not draining connections RDS_PROXY_POOL_DRAIN describe-db-proxies; CloudWatch DatabaseConnections metric on the proxy
Recovery mode causes long failover RECOVERY_MODE_FULL Check parameter group for recovery_mode or Aurora parallel recovery settings
No CloudWatch event for failover CLOUDWATCH_EVENT_GAP describe-events for the cluster; check EventBridge rule for RDS events

Pre-flight: gather the failover context

Before running root-cause-specific probes, gather the canonical cluster configuration and short-circuit on misconfigurations that mimic failover failures.

Account-wide pre-flight commands

Moved verbatim to references/diagnostic-commands.md — load on demand.

Cluster-state short-circuit

Finding Effect on diagnosis
Cluster Status: available, writer endpoint resolves to a healthy instance Failover completed at the DB layer. Investigate application-side issues (endpoint, connection pool).
Cluster Status: modifying or Status: upgrading A maintenance action is in progress. Failover may be delayed or blocked. Wait for completion.
Instance Status: storage-full The instance has exhausted allocated storage. Failover cannot complete. ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: STORAGE_FULL.
Instance MultiAZ: false The instance is not Multi-AZ enabled. No standby exists. Failover cannot occur.
Instance PromotionTier on intended target > 0 A lower-tier instance will be promoted first. Check if another instance has tier 0.
Parameter group status: applying or pending-reboot A parameter change is pending. The standby may have a different effective configuration.
Aurora Global cluster with GlobalClusterIdentifier set The cluster is part of a Global DB. Failover semantics differ (managed vs unmanaged).

Moved verbatim to references/error-handling.md — load on demand.

Process — Diagnostic decision tree (apply in symptom order)

The diagnostic tree is symptom-driven. Pick the entry point based on the observed symptom, then walk the root-cause-specific probes in order.

Step 0: Non-obvious behaviours that change diagnosis

Moved verbatim to references/advanced-patterns.md — load on demand.

Step 1: Application connection string check

Symptom: failover completed (new writer is active, cluster endpoint resolves correctly), but the application still cannot connect.

Moved verbatim to references/diagnostic-commands.md — load on demand.

If the application uses the instance endpoint (<cluster>-instance-1.<id>.<region>.rds.amazonaws.com) instead of the cluster endpoint (<cluster>.cluster-<id>.<region>.rds.amazonaws.com), ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: APP_NOT_CLUSTER_ENDPOINT.

The instance endpoint does not follow the writer after failover. The application must use the cluster endpoint (also called the writer endpoint) for transparent failover.

Step 2: DNS propagation check

Symptom: application uses the cluster endpoint but still hits the old instance for 30+ seconds after failover.

Moved verbatim to references/diagnostic-commands.md — load on demand.

Check the application's DNS caching layers:

Moved verbatim to references/error-handling.md — load on demand.

If a caching layer is holding the old IP, ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: DNS_PROPAGATION_DELAY (OS/JVM DNS cache) or ROOT_CAUSE: CONNECTION_POOL_CACHING (connection pool).

Step 3: Multi-AZ health check threshold

Symptom: primary instance is unhealthy (high CPU, stuck queries) but failover does not trigger.

Moved verbatim to references/diagnostic-commands.md — load on demand.

Multi-AZ failover triggers when:

  • The primary instance becomes unreachable (network partition).
  • The instance status changes to failed.
  • The storage subsystem becomes unresponsive.

Multi-AZ does NOT trigger on:

  • High CPU (even at 100%).
  • High connection count.
  • Slow queries.

If the primary is degraded but responsive (high CPU, slow queries), Multi-AZ will not failover. ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: MULTI_AZ_HEALTH_THRESHOLD. The health check only triggers on unreachability, not performance degradation.

Step 4: Failover priority tier

Symptom: Aurora failover completed, but the "wrong" instance became the new writer.

Moved verbatim to references/diagnostic-commands.md — load on demand.

Aurora failover promotion order:

  1. Lowest PromotionTier (tier 0 first).
  2. Within the same tier, largest DBInstanceClass.
  3. Within the same tier and class, longest uptime.

If the intended promotion target has tier 15 (default) but another instance has tier 0, the tier 0 instance is promoted. ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: FAILOVER_PRIORITY_TIER.

Fix: set the intended target to tier 0:

Moved verbatim to references/diagnostic-commands.md — load on demand.

Step 5: Storage-full check

Symptom: failover event appears in RDS events but the new writer does not become available. Error: InsufficientStorage or FailoverFailed.

Moved verbatim to references/diagnostic-commands.md — load on demand.

If FreeStorageSpace is near zero (< 5 GB) or DBInstanceStatus is storage-full, ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: STORAGE_FULL.

Fix: increase allocated storage:

Moved verbatim to references/diagnostic-commands.md — load on demand.

Step 6: Parameter group mismatch

Symptom: standby instance fails health checks during promotion. RDS events show a promotion failure with parameter-related errors.

Moved verbatim to references/diagnostic-commands.md — load on demand.

If the primary and standby have different DB parameter groups with incompatible settings, the standby may fail during promotion. ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: PARAMETER_GROUP_MISMATCH.

Fix: align the parameter groups:

Moved verbatim to references/diagnostic-commands.md — load on demand.

Step 7: Option group conflict

Symptom: standby cannot start the engine after promotion. RDS events show an option-group-related error.

Moved verbatim to references/diagnostic-commands.md — load on demand.

If the option groups are incompatible (e.g., primary has TDE, standby does not), ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: OPTION_GROUP_CONFLICT.

Fix: align the option groups:

Moved verbatim to references/diagnostic-commands.md — load on demand.

Step 8: Read replica promotion failure

Symptom: a read replica (Aurora or cross-region) cannot be promoted to a standalone cluster/instance.

Moved verbatim to references/diagnostic-commands.md — load on demand.

Common promotion failures:

  • High replication lag (the replica has not caught up).
  • Network connectivity to the source (for cross-region).
  • Storage-full on the replica.
  • Parameter group mismatch.

If any of these apply, ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: READ_REPLICA_PROMOTION_FAILURE.

Step 9: Aurora Global DB failover mode

Symptom: Aurora Global DB failover did not work or took too long.

Moved verbatim to references/diagnostic-commands.md — load on demand.

Managed failover (failover-global-cluster) is a control-plane operation that takes 1-5 minutes. Unmanaged failover (detach secondary

  • promote) is faster but requires manual coordination.

If the failover mode is misconfigured or the secondary cluster has high replication lag, ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: AURORA_GLOBAL_FAILOVER_MODE.

Step 10: RDS Proxy failover

Symptom: RDS Proxy does not transparently handle failover; application sees connection errors.

Moved verbatim to references/diagnostic-commands.md — load on demand.

RDS Proxy should queue and reconnect during failover. If connections are being dropped instead of queued, check:

  • Proxy target group health.
  • ConnectionBorrowTimeout (client-side).
  • Proxy IdleClientTimeout.

If the Proxy is not draining/queuing correctly, ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: RDS_PROXY_POOL_DRAIN.

Step 11: Recovery mode

Symptom: Aurora failover takes 30+ seconds consistently; Multi-AZ failover takes 120+ seconds.

For Aurora, check if parallel recovery is enabled (engine-version dependent):

Moved verbatim to references/diagnostic-commands.md — load on demand.

Aurora PostgreSQL 13+ supports parallel WAL replay. Aurora MySQL 8.0 has optimised crash recovery. Older versions use serial recovery.

For Multi-AZ, check the recovery_mode parameter:

Moved verbatim to references/diagnostic-commands.md — load on demand.

If recovery mode is set to full instead of optimized, ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: RECOVERY_MODE_FULL.

Step 12: CloudWatch failover event gap

Symptom: failover occurred but no CloudWatch alarm or EventBridge rule fired.

Moved verbatim to references/diagnostic-commands.md — load on demand.

Moved verbatim to references/error-handling.md — load on demand.

If no events appear in describe-events, the failover may not have triggered at the RDS service level. If events appear but no alarm fired, the EventBridge rule or CloudWatch alarm is misconfigured. ROOT_CAUSE_IDENTIFIED with ROOT_CAUSE: CLOUDWATCH_EVENT_GAP.

Output format

TARGET: <cluster-id or instance-id, application context>
VERDICT: ROOT_CAUSE_IDENTIFIED | INSUFFICIENT_DATA
REASON: <1-2 sentences naming the root cause and the failing probe>
ROOT_CAUSE: <APP_NOT_CLUSTER_ENDPOINT | MULTI_AZ_HEALTH_THRESHOLD |
              FAILOVER_PRIORITY_TIER | STORAGE_FULL |
              PARAMETER_GROUP_MISMATCH | OPTION_GROUP_CONFLICT |
              DNS_PROPAGATION_DELAY | CONNECTION_POOL_CACHING |
              AURORA_GLOBAL_FAILOVER_MODE | RDS_PROXY_POOL_DRAIN |
              RECOVERY_MODE_FULL | CLOUDWATCH_EVENT_GAP |
              READ_REPLICA_PROMOTION_FAILURE | UNKNOWN>
EVIDENCE:
  - <observed symptom — error string, failover timeline>
  - <failing probe — command and its output that confirms the cause>
  - <passing probes — layers ruled out>
REMEDIATION:
  1. <specific action with CLI command>
  2. <verification command after the fix>
CONFIRM: Before executing any state-changing CLI, emit and await operator
  approval: "CONFIRM: About to <action> on <cluster> in <region>. Proceed?
  (yes/no)"

Worked example — Application using instance endpoint

TARGET: prod-db-cluster, application payments-service
VERDICT: ROOT_CAUSE_IDENTIFIED
REASON: The Aurora cluster failover completed at 03:17 UTC (confirmed
  via RDS events). The writer cluster endpoint resolves to the new
  writer instance within 1 second (DNS TTL verified). However, the
  application connection string uses the instance endpoint
  (prod-db-cluster-instance-1.xxxx.us-east-1.rds.amazonaws.com)
  instead of the cluster endpoint. After failover, instance-1 became
  a reader; the application is connecting to a reader and receiving
  "cannot execute WRITE in a read-only transaction" errors.
ROOT_CAUSE: APP_NOT_CLUSTER_ENDPOINT
EVIDENCE:
  - Symptom: application started receiving "read-only transaction"
    errors at 03:17 UTC, the same time as the Aurora failover event.
  - Probe: aws rds describe-db-clusters shows the cluster Status is
    "available"; the writer endpoint resolves to instance-2 (the new
    writer).
  - Probe: application config shows connection string:
    jdbc:postgresql://prod-db-cluster-instance-1.xxxx.us-east-1.rds.amazonaws.com:5432/prod_db
    (instance endpoint, NOT cluster endpoint).
  - Probe: dig prod-db-cluster-instance-1.xxxx resolves to the reader
    IP; dig prod-db-cluster.cluster-xxxx resolves to the writer IP.
  - Passing: DNS TTL is 1 second (not a DNS issue); connection pool
    maxLifetime is 30s (not a pool caching issue).
REMEDIATION:
  1. Update the application connection string to use the cluster
     endpoint:
     jdbc:postgresql://prod-db-cluster.cluster-xxxx.us-east-1.rds.amazonaws.com:5432/prod_db
  2. Restart or reload the application to pick up the new connection
     string.
  3. Verify by triggering a test failover and confirming the
     application reconnects to the new writer within 30 seconds.
CONFIRM: Before updating the application configuration, confirm:
  "CONFIRM: About to update the payments-service connection string
   from instance endpoint to cluster endpoint. Proceed? (yes/no)"

Worked example — Connection pool caching

Moved verbatim to references/worked-examples.md — load on demand.

Worked example — INSUFFICIENT_DATA

Moved verbatim to references/worked-examples.md — load on demand.

Anti-Patterns — NEVER

  • NEVER declare ROOT_CAUSE_IDENTIFIED without a failing probe that matches the symptom. A "process of elimination" diagnosis erodes operator trust when the real cause is elsewhere.

  • NEVER recommend using the instance endpoint for production application connections. The instance endpoint does not follow the writer after failover. Always use the cluster endpoint (writer endpoint for writes, reader endpoint for reads).

  • NEVER assume the failover mechanism is broken because the application is down. Aurora and Multi-AZ failover automation works correctly in the vast majority of cases. The root cause is usually in the application layer (wrong endpoint, stale pool) or configuration (tier, storage, parameter group).

  • NEVER trigger a manual failover to "test" without confirming the application will handle it. If the application uses the instance endpoint or has stale connection pools, a test failover will cause an outage.

  • NEVER conclude "failover took 5 minutes" without checking the database failover time separately from the application recovery time. The DB failover may have completed in 10 seconds; the application's connection pool held stale connections for 4.5 minutes.

  • NEVER set PromotionTier to 0 on all instances without understanding the implication. When multiple instances have tier 0, the largest is chosen. If you want a specific instance to be the promotion target, set it to tier 0 and all others to tier 1+.

  • NEVER ignore FreeStorageSpace when troubleshooting failover. A storage-full instance cannot complete failover. The error does not say "storage full" — it says FailoverFailed or InsufficientStorage.

  • NEVER assume Aurora Global DB failover is as fast as single-region failover. Managed Global failover takes 1-5 minutes (control-plane operation + cross-region DNS). Plan for this latency in your RTO.

  • NEVER use recovery_mode: full when optimized is available. Full recovery mode extends failover duration by 30-60 seconds for no benefit in production workloads.

  • NEVER assume RDS Proxy eliminates all failover impact. The Proxy reduces connection errors but active transactions on the old writer are rolled back. The Proxy pool drains over 5-30 seconds.

  • NEVER modify parameter groups or option groups on a production cluster without verifying compatibility between primary and standby. A mismatch can prevent the standby from being promoted.

  • NEVER conclude no failover occurred just because no CloudWatch alarm fired. Check describe-events directly — the alarm may be misconfigured, not the failover.

Pre-flight safety checks (run before any state-changing CLI)

  • MANDATORY CONFIRMATION GATE. Before any state-changing operation (failover-db-cluster, modify-db-instance, promote-read-replica, failover-global-cluster), emit and await operator approval.

  • Read-only first. Every probe in the diagnostic tree is read-only (describe-db-clusters, describe-db-instances, describe-events, get-metric-statistics, dig). Do not perform state-changing operations as diagnostic probes.

  • Manual failover causes a brief outage (10-120 seconds depending on engine). Always confirm before failover-db-cluster.

  • modify-db-instance --promotion-tier is applied immediately but does not affect the current writer. It only affects the next failover.

  • Increasing allocated storage (--allocated-storage) is applied online for most engines but may cause a brief I/O suspension for older MySQL versions.

  • Parameter group changes may require a reboot to take effect. For Multi-AZ, the reboot triggers a failover. Always use --no-apply-immediately if the change should wait for the next maintenance window.

  • Option group changes can cause the engine to restart. For Multi-AZ, this triggers a failover.

Remediation guidance

Moved verbatim to references/error-handling.md — load on demand.

Deep reference: Aurora and Multi-AZ failover model

Moved verbatim to references/aurora-endpoint-reference.md — load on demand.

Moved verbatim to references/failover-priority-reference.md — load on demand.

Moved verbatim to references/advanced-patterns.md — load on demand.

Recent AWS features (2024-2026)

Moved verbatim to references/advanced-patterns.md — load on demand.

References (load on demand)

Domain

AWS CloudOps / RDS and Aurora Database Failover, High Availability, Cluster Endpoint Management, Multi-AZ Configuration, Aurora Global DB, and RDS Proxy.

AWS documentation

Moved verbatim to references/advanced-patterns.md — load on demand.

04 同仓库更多 Skills

jhkchan
jhkchan/aws-agent-skills

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 / method / 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 / 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 / EKS / 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.

★ 4
jhkchan
jhkchan/aws-agent-skills

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.

★ 4
jhkchan
jhkchan/aws-agent-skills

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 / 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.

★ 4
jhkchan
jhkchan/aws-agent-skills

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/API Gateway/CloudFront, or configuring CAPTCHA/Challenge 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.

★ 4
jhkchan
jhkchan/aws-agent-skills

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/bypass 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.

★ 4
jhkchan
jhkchan/aws-agent-skills

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.

★ 4