Skip to content

borghei/Claude-Skills / engineering/senior-devops

senior-devops

Verified

DevOps for CI/CD, containers, Kubernetes, and Terraform. Use when building pipelines, containerizing apps, managing clusters, provisioning cloud infra, deploying with blue- green/canary, or handling infrastructure incidents.

references/cloud_platform_guide.md

View on GitHub
# Cloud Platform Guide

## Service Comparison Matrix

| Capability | AWS | GCP | Azure |
|------------|-----|-----|-------|
| **Managed K8s** | EKS | GKE | AKS |
| **Serverless** | Lambda | Cloud Functions / Cloud Run | Azure Functions |
| **Containers** | ECS/Fargate | Cloud Run | Container Apps |
| **Object Storage** | S3 | Cloud Storage | Blob Storage |
| **Managed DB** | RDS / Aurora | Cloud SQL / AlloyDB | Azure SQL / Cosmos DB |
| **Message Queue** | SQS / SNS | Pub/Sub | Service Bus |
| **CDN** | CloudFront | Cloud CDN | Azure CDN / Front Door |
| **DNS** | Route 53 | Cloud DNS | Azure DNS |
| **Secrets** | Secrets Manager | Secret Manager | Key Vault |
| **IAM** | IAM + STS | IAM + Workload Identity | Entra ID + RBAC |

## Multi-Cloud Decision Framework

**When multi-cloud makes sense:**
- Regulatory requirements mandate vendor diversity
- Acquisition brings workloads on a different cloud
- Best-of-breed services (GCP for ML, AWS for breadth)

**When it does not:**
- Avoiding lock-in as the sole motivation (operational tax exceeds savings)
- Small teams that cannot afford complexity overhead

**If you go multi-cloud:**
- Use Terraform for the abstraction layer
- Standardize on Kubernetes as compute plane
- Centralize observability (Datadog, Grafana Cloud)
- Invest in a platform engineering team

## Cost Optimization

### Right-Sizing Methodology

1. Collect 2-4 weeks of CPU/memory utilization
2. Identify instances below 40% average CPU
3. Recommend one size down
4. Validate in staging under load test
5. Apply in production during maintenance window
6. Monitor for 1 week post-change

### Spot/Preemptible Strategy

| Workload | Spot? | Pattern |
|----------|-------|---------|
| Stateless web (behind LB) | Yes | 70% spot + 30% on-demand |
| CI/CD runners | Yes | 100% spot with retry |
| Batch / ETL | Yes | Spot fleet with checkpointing |
| Databases / stateful | No | Reserved instances |
| Dev/test environments | Yes | 100% spot |

### FinOps Practices

- **Tagging**: Enforce `team`, `environment`, `service`, `cost-center` on all resources
- **Budget alerts**: 50%, 80%, 100% of monthly budget
- **Reserved capacity**: 1-year for baseline workloads (30-40% savings)
- **Scheduled scaling**: Scale down non-prod outside business hours
- **Storage lifecycle**: S3 lifecycle policies for Glacier/Archive tiers
- **Unused resources**: Weekly scan for unattached EBS, idle LBs, stale snapshots

## Prometheus Alerting Rules

```yaml
groups:
  - name: application
    rules:
      - alert: HighErrorRate
        expr: sum(rate(http_requests_total{status=~"5.."}[5m])) / sum(rate(http_requests_total[5m])) > 0.05
        for: 5m
        labels: { severity: critical }
        annotations:
          summary: "Error rate exceeds 5% for 5 minutes"

      - alert: HighLatencyP99
        expr: histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket[5m])) by (le)) > 2.0
        for: 10m
        labels: { severity: warning }

      - alert: PodCrashLooping
        expr: increase(kube_pod_container_status_restarts_total[1h]) > 5
        for: 5m
        labels: { severity: critical }

      - alert: DiskSpaceLow
        expr: (node_filesystem_avail_bytes{mountpoint="/"} / node_filesystem_size_bytes{mountpoint="/"}) < 0.15
        for: 10m
        labels: { severity: warning }
```

## Incident Response Quick Reference

### Severity Classification

| Severity | Definition | Response Time |
|----------|-----------|---------------|
| SEV-1 | Complete outage, data loss risk | 15 min |
| SEV-2 | Significant degradation | 30 min |
| SEV-3 | Minor degradation, workaround available | 4 hours |
| SEV-4 | Cosmetic / informational | Next business day |

### Runbook Template

```markdown
## Symptoms
- What alerts fire, what users report

## Diagnosis
1. kubectl get pods -n production -l app=myapp
2. helm history myapp -n production
3. kubectl logs -l app=myapp --tail=100

## Quick Fix (< 5 min)
- kubectl rollout restart deployment/myapp
- kubectl scale deployment/myapp --replicas=10

## Rollback (< 10 min)
- helm rollback myapp [previous-revision]
```

### Postmortem (required for SEV-1/SEV-2)
1. Timeline reconstruction
2. Root cause (5 Whys)
3. Impact assessment
4. What went well / poorly
5. Action items with owners and due dates
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 197kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall