mojahid0862/devops-doctor

Diagnose AWS deployment failures and production incidents with read-only evidence first across ECS/Fargate, EC2, Lambda, ALB/NLB, RDS, ElastiCache/Redis, S3, CloudFront, Route 53, IAM, VPC/security groups, CloudWatch, CloudTrail, GitLab deploy jobs, and rollback paths. Use when the user reports an AWS deploy failure, outage, unhealthy target, failed release, or cloud incident.

Find AWS cost waste and rightsizing opportunities using read-only Cost Explorer, CloudWatch, EC2, ECS, RDS, ElastiCache/Redis, EBS, EIP, NAT gateway, load balancer, S3, CloudFront, Route 53, CloudWatch Logs, and ECR evidence. Use when the user asks to reduce AWS cost, find waste, optimize spend, or check over-provisioning.

Route DevOps tasks across code review, GitLab delivery, AWS incidents and health, infrastructure cost, security, and Terraform. Apply shared task assessment and adaptive delegation across domains, then select the relevant specialist.

Coordinate a multi-step GitLab and AWS investigation from chat, selecting scoped evidence collectors and tracking unresolved questions. Use when the user asks for end-to-end triage or delegated operations across multiple services.

Coordinate adaptive subagents across DevOps Doctor tasks: coding, reviews, CI/CD, cost, security, IaC, incidents, and health checks. Use whenever independent work or review benefits from delegation, without requiring an explicit agents request. Keep small or strictly dependent work direct.

Diagnose ECS and Fargate service failures, stopped tasks, unhealthy ALB targets, image pull errors, task definition drift, capacity provider issues, CPU/memory pressure, CloudWatch logs, IAM execution role problems, and deployment rollbacks. Use when the user mentions ECS, Fargate, task stopped, target health, container exit, or service not stable.

Review a GitLab MR using glab-only metadata, the actual diff, and current-commit CI evidence. Use for correctness, security, performance, delivery, or production-risk review before merge; route failed-job root-cause requests to gitlab-pipeline-doctor.

Investigate GitLab CI/CD pipeline failures, failed jobs, runner errors, job traces, artifacts, cache, Docker build failures, deploy gates, and GitLab pipeline or job URLs using glab-only read-only evidence. Use when the user gives a failed GitLab pipeline/job URL, red CI status, job log, runner failure, or asks why GitLab CI failed.

Diagnose incidents using logs, metrics, traces, alerts, SLOs, CloudWatch, Grafana, New Relic, ALB target health, CloudFront/Route 53 signals, ElastiCache/Redis health, ECS events, Kubernetes events, application logs, and deployment timelines. Use when the user gives an outage, alert, latency spike, error spike, noisy monitor, missing dashboard, or asks what to check in observability.

Create safe rollback and recovery plans for GitLab deployments, AWS ECS/Fargate, EC2, Lambda, RDS, ElastiCache/Redis, Kubernetes, Terraform, Docker, migrations, feature flags, CloudFront/Route 53 DNS/CDN, and failed releases. Use when the user asks how to roll back, reduce blast radius, recover production, or make a deployment rollback-safe.

Check cloud, GitLab CI/CD, Docker, Kubernetes, Terraform, IAM, S3, CloudFront, Route 53, ElastiCache/Redis, security groups, secrets handling, public exposure, dependency supply chain, and observability security drift without exposing secret values. Use when the user asks for DevOps security review, drift check, hardening, OWASP/cloud risk, or safe production posture.

Review Terraform plans and IaC diffs for destructive changes, replacements, IAM privilege creep, public exposure, missing encryption, deletion protection, unsafe security groups, cost spikes, tagging gaps, drift, and rollout safety. Use when the user shares Terraform code, a terraform plan, plan JSON, or asks if an infrastructure change is safe.