It's 3 AM. Your phone buzzes — a PagerDuty alert. A pod is in CrashLoopBackOff, and you're the only one awake. You stumble to your laptop, open a terminal, and start typing kubectl describe pod. Sound familiar? According to a 2023 survey by Dimensional Research, 62% of DevOps engineers report being woken up at least once a week for issues that could have been resolved during business hours. But what if you had an AI assistant that could guide you through the fix in seconds?
This roundup is not about replacing your team — it's about augmenting your on-call shifts with 12 battle-tested prompts that cover the most common Kubernetes, Docker, and CI/CD pain points. Each prompt includes a real-world scenario, the exact prompt text, and the expected output. We've validated these against real clusters running on AWS EKS and Docker Swarm, so you can copy-paste with confidence.
1. Decoding CrashLoopBackOff: From Logs to Root Cause
Scenario: A deployment suddenly starts crashing. You have logs but need to identify the exact reason.
Prompt:
You are a Kubernetes troubleshooting expert. Analyze the following pod events and container logs. Identify the root cause and suggest a fix. Use the format: [Cause] -> [Evidence] -> [Fix].
Events:
...
Logs:
...
Example:
Events show Back-off restarting failed container and logs show connection refused to database at 10.0.0.5:5432.
Output:
- Cause: Database connection refused.
- Evidence: Log line connection refused.
- Fix: Check if the database service is running and if network policies allow traffic from the pod's namespace. Use kubectl get svc -n db and kubectl describe networkpolicy.
Why it works: The AI correlates events and logs, saving you from manual grep.
2. Dockerfile Optimization: Shrinking Images by 70%
Scenario: Your Node.js image is 1.2GB. You want to reduce it.
Prompt:
Optimize the following Dockerfile for a Node.js application. Use multi-stage builds, alpine base images, and remove unnecessary dependencies. Provide the optimized Dockerfile and explain each change.
Example:
Original Dockerfile uses node:14 and copies everything.
Output: Multi-stage build with node:14-alpine for builder and node:14-alpine for runtime, copying only node_modules and dist. Result: 350MB.
Why it works: AI suggests best practices like npm ci --only=production and .dockerignore.
3. CI/CD Pipeline Debugging: When GitHub Actions Fail
Scenario: Your GitHub Actions workflow fails at the test step with a cryptic error.
Prompt:
You are a CI/CD expert. Given the following GitHub Actions workflow YAML and error log, identify the failure and suggest a fix. Focus on environment variables, caching, and dependency versions.
Workflow:
...
Error:
...
Example:
Error: Error: Cannot find module 'jest'.
Output: The workflow uses npm install but the cache is stale. Fix: Add npm ci and configure actions/cache for node_modules.
Why it works: AI spots missing steps and version mismatches.
4. Kubernetes Resource Limits: Avoiding OOMKills
Scenario: A pod keeps getting OOMKilled. You need to set proper requests and limits.
Prompt:
Analyze the following pod specification and metrics. Recommend CPU and memory requests/limits to prevent OOMKills while maintaining performance. Use the format: [Current] -> [Recommended] -> [Reason].
Pod spec:
...
Metrics (from kubectl top):
...
Example:
Current: No limits. Metrics: 200m CPU, 512Mi memory. Recommended: requests: 250m CPU, 600Mi memory; limits: 500m CPU, 1Gi memory.
Why it works: AI uses real metrics to suggest values, avoiding guesswork.
5. Helm Chart Generation from Scratch
Scenario: You need to create a Helm chart for a new microservice.
Prompt:
Generate a Helm chart for a Node.js microservice with the following requirements: deployment with 3 replicas, service on port 3000, ingress with TLS, and configmap for environment variables. Output the chart structure and key files.
Example:
Output includes Chart.yaml, values.yaml, templates/deployment.yaml, etc.
Why it works: AI produces boilerplate that you can customize, saving hours.
6. Docker Compose for Local Development
Scenario: You need a local environment with PostgreSQL, Redis, and your app.
Prompt:
Write a docker-compose.yml for a Node.js app that depends on PostgreSQL and Redis. Include healthchecks, volumes for data persistence, and environment variables. Ensure services start in the correct order.
Example:
Output includes depends_on with condition: service_healthy and healthchecks for db and redis.
Why it works: AI knows the correct syntax for healthchecks and dependencies.
7. Kubernetes Network Policy Troubleshooting
Scenario: Pods can't communicate across namespaces.
Prompt:
Given the following network policies and pod labels, explain why traffic from namespace A to namespace B is blocked. Suggest a corrected policy.
Policies:
...
Pod labels:
...
Example:
Policy denies all ingress; need to add allow rule for specific labels.
Why it works: AI parses YAML and identifies missing rules.
8. CI/CD Security: Scanning for Vulnerabilities
Scenario: You want to integrate Trivy into your pipeline.
Prompt:
Add a Trivy scan step to the following GitHub Actions workflow. The scan should run on every push to main and fail if HIGH or CRITICAL vulnerabilities are found. Provide the updated YAML.
Example:
Output adds a step using aquasecurity/trivy-action with exit-code: 1 and severity: HIGH,CRITICAL.
Why it works: AI knows the action's parameters.
9. Kubernetes Autoscaling: HPA Configuration
Scenario: Your app experiences traffic spikes. You need to set up HPA.
Prompt:
Create a HorizontalPodAutoscaler for a deployment named 'web' that scales between 2 and 10 replicas based on CPU utilization of 70%. Include the YAML and explain how it works.
Example:
Output includes apiVersion: autoscaling/v2, metrics: - type: Resource, resource: name: cpu, target: averageUtilization: 70.
Why it works: AI uses the correct API version and syntax.
10. Docker Volume Backup and Restore
Scenario: You need to backup a database volume regularly.
Prompt:
Write a bash script that backs up a Docker volume named 'db_data' to a timestamped tar file and restores it. Include error handling.
Example:
Script uses docker run --rm -v db_data:/data -v $(pwd):/backup alpine tar czf /backup/backup.tar.gz -C /data ..
Why it works: AI provides a working script with comments.
11. Kubernetes Ingress with Cert-Manager
Scenario: You need to expose a service with automatic TLS.
Prompt:
Create an Ingress resource for a service 'web' on domain example.com with TLS using cert-manager. Use the cluster-issuer 'letsencrypt-prod'. Provide the YAML.
Example:
Output includes annotations cert-manager.io/cluster-issuer: letsencrypt-prod and tls: - hosts: [example.com], secretName: example-tls.
Why it works: AI knows the cert-manager annotations.
12. CI/CD Pipeline for Multi-Environment Deployment
Scenario: You need a pipeline that deploys to staging and production with manual approval.
Prompt:
Write a GitHub Actions workflow that builds a Docker image, pushes to ECR, and deploys to staging automatically, then to production after manual approval. Use environment secrets.
Example:
Output includes jobs for build, deploy-staging, and deploy-production with environment: production and needs: deploy-staging.
Why it works: AI structures the workflow correctly with dependencies.
These 12 prompts are just the tip of the iceberg. The key is to treat AI as a junior engineer who never sleeps — it can draft, debug, and document while you focus on architecture. Start with one prompt today, and watch your on-call burden shrink. For more advanced AI-driven DevOps workflows, explore our courses at asibiont.com.
Comments