Helm Scaffold

Helm Scaffold

Automate production-ready Helm chart generation using template assets, Python scaffolding scripts, and Kubernetes best practices through conversational interaction.

Core Capabilities

• Automated scaffolding using Python script (scripts/scaffold_chart.py)
• Template-based generation from assets/templates/ directory
• Convert Kubernetes manifests to Helm templates
• Multi-environment configurations (dev, staging, prod)
• Organizational standardization with team policies
• Comprehensive testing with dry-run workflows

Quick Start

For simple chart generation, use the Python scaffolding script directly:

python3 scripts/scaffold_chart.py <chart-name> \
  --workload-type deployment \
  --output /mnt/user-data/outputs \
  --ingress \
  --hpa

Workflow Decision Tree

User Request
├─ Simple chart (name + type provided)
│  └─ Use scaffold_chart.py directly
├─ Complex chart (many options)
│  └─ Ask clarifying questions, then use scaffold_chart.py
├─ Manifest conversion
│  └─ Manual templatization process
└─ Custom requirements
   └─ Manual generation with template references

Interactive Workflow

Step 1: Understand Use Case

Identify the scenario:

• Simple new chart: Use scaffold_chart.py with user-provided params
• Complex chart: Ask questions, build command
• Manifest conversion: Extract variables, templatize
• Team template: Apply organizational standards

Step 2: Gather Requirements

Minimum required:

• Chart name
• Workload type (deployment, statefulset, job, cronjob)

Optional (ask if not provided):

• Container image repository and tag
• Application port
• Include Ingress? (--ingress flag)
• Include HPA? (--hpa flag)
• Include ConfigMap? (--configmap flag)
• Multi-environment configs?

Step 3: Generate Chart

Option A: Using scaffold_chart.py (Recommended)

For straightforward charts, execute the Python script:

import subprocess
cmd = [
    'python3', 'scripts/scaffold_chart.py',
    chart_name,
    '--workload-type', workload_type,
    '--output', '/mnt/user-data/outputs'
]
if include_ingress:
    cmd.append('--ingress')
if include_hpa:
    cmd.append('--hpa')
if include_configmap:
    cmd.append('--configmap')

subprocess.run(cmd, check=True)

The script automatically:

• Creates chart directory structure
• Copies template files from assets/templates/
• Replaces CHARTNAME placeholder with actual chart name
• Includes only requested resources
• Applies best practices

Option B: Manual Generation

For custom requirements, manually copy and modify templates:

1. Read template from assets/templates/<workload>/<file>.yaml
2. Replace CHARTNAME with actual chart name
3. Customize based on user requirements
4. Write to /home/claude/<chart-name>/templates/

Step 4: Multi-Environment Configuration

If user requests dev/staging/prod configs, create additional values files:

values-dev.yaml:

replicaCount: 1
resources:
  limits: {cpu: 200m, memory: 128Mi}
  requests: {cpu: 25m, memory: 32Mi}
env:
  - name: LOG_LEVEL
    value: "debug"

values-prod.yaml:

replicaCount: 3
resources:
  limits: {cpu: 1000m, memory: 512Mi}
  requests: {cpu: 100m, memory: 128Mi}
autoscaling:
  enabled: true
  minReplicas: 3
  maxReplicas: 10

Step 5: Testing Instructions

Always provide comprehensive testing workflow:

# Navigate to chart directory
cd <chart-name>

# 1. Validate chart structure
helm lint .

# 2. Render templates locally
helm template <chart-name> .

# 3. Dry run installation
helm install <chart-name> . --dry-run --debug

# 4. Test with specific values
helm template <chart-name> . -f values-dev.yaml

# 5. Deploy to test namespace
kubectl create namespace test
helm install <chart-name> . -n test

# 6. Verify deployment
kubectl get all -n test
helm status <chart-name> -n test

# 7. Cleanup
helm uninstall <chart-name> -n test
kubectl delete namespace test

Step 6: Deliver Output

• Chart is created in /mnt/user-data/outputs/<chart-name>/
• Provide download link
• Include testing instructions
• Explain customization options

Workload Types

Load references/workload-types.md for detailed decision tree and characteristics.

Quick reference:

• Deployment: Stateless apps (web, API, microservices)
• StatefulSet: Stateful apps (databases, caches) - stable IDs, persistent storage
• Job: One-time tasks (migrations, ETL)
• CronJob: Scheduled tasks (backups, reports)

Template locations:

• assets/templates/deployment/deployment.yaml
• assets/templates/statefulset/statefulset.yaml
• assets/templates/job/job.yaml
• assets/templates/cronjob/cronjob.yaml

Converting Manifests to Helm

When user provides raw Kubernetes YAML:

1. Analyze manifests: Identify resources and configurable values
2. Extract variables: Images, replicas, ports, resources, env-specific settings
3. Create values.yaml: Organize extracted values logically
4. Templatize YAML:
   • Replace hardcoded values with {{ .Values.* }}
   • Use {{ include "CHARTNAME.fullname" . }} for names
   • Use {{ include "CHARTNAME.labels" . }} for labels
5. Add helpers: Copy assets/templates/_helpers.tpl and customize
6. Document: Explain what was parameterized

Template Assets Structure

assets/templates/
├── Chart.yaml           # Base chart metadata
├── values.yaml          # Complete values with all options
├── .helmignore          # Files to ignore
├── _helpers.tpl         # Helper functions (CHARTNAME placeholder)
├── NOTES.txt            # Post-install instructions
├── deployment/
│   └── deployment.yaml  # Deployment template
├── statefulset/
│   └── statefulset.yaml # StatefulSet template
├── job/
│   └── job.yaml         # Job template
├── cronjob/
│   └── cronjob.yaml     # CronJob template
├── service/
│   └── service.yaml     # Service template
├── ingress/
│   └── ingress.yaml     # Ingress template
├── hpa/
│   └── hpa.yaml         # HPA template
├── configmap/
│   └── configmap.yaml   # ConfigMap template
└── rbac/
    └── serviceaccount.yaml  # ServiceAccount template

All templates use CHARTNAME placeholder which is replaced by the script.

Scripts

scaffold_chart.py

Purpose: Automated chart generation from templates

Usage:

python3 scripts/scaffold_chart.py CHART_NAME [OPTIONS]

Arguments:
  CHART_NAME              Name of the Helm chart

Options:
  -w, --workload-type    Type: deployment, statefulset, job, cronjob (default: deployment)
  -o, --output          Output directory (default: current directory)
  --ingress             Include Ingress resource
  --hpa                 Include HorizontalPodAutoscaler
  --configmap           Include ConfigMap

What it does:

• Creates chart directory structure
• Copies relevant templates from assets/templates/
• Replaces CHARTNAME placeholder
• Includes only requested optional resources
• Applies best practices automatically

Best Practices (Auto-Applied)

The templates in assets/templates/ already include:

• ✅ Standard Kubernetes labels (app.kubernetes.io/*)
• ✅ Security contexts (readOnlyRootFilesystem, runAsNonRoot, dropped capabilities)
• ✅ Resource limits and requests
• ✅ Health checks (liveness and readiness probes)
• ✅ Helper functions for naming and labels
• ✅ Proper selector labels
• ✅ Service account configuration

Organizational Standardization

For platform teams needing consistent charts:

1. Capture standards: Ask about required labels, policies, security requirements
2. Modify templates: Update assets/templates/_helpers.tpl with org labels
3. Generate: Use scaffold_chart.py with modified templates
4. Document: Explain customization points

Example org-specific helper addition:

{{- define "CHARTNAME.orgLabels" -}}
org.example.com/cost-center: {{ .Values.org.costCenter | required "Cost center required" }}
org.example.com/team: {{ .Values.org.team | required "Team name required" }}
{{- end }}

Success Criteria

Generated charts must:

• ✅ Pass helm lint without errors (95%+ target)
• ✅ Follow CNCF/Helm best practices
• ✅ Include security contexts by default
• ✅ Have resource limits configured
• ✅ Use standard Kubernetes labels
• ✅ Be deployment-ready within <15 minutes

References

Load as needed based on use case:

For workload selection:

• references/workload-types.md - Decision tree, characteristics, when to use each type

For testing:

• references/testing-guide.md - Comprehensive dry-run workflows, mock values, validation commands

For best practices:

• references/best-practices.md - CNCF/Helm standards, security guidelines

For examples:

• references/examples.md - Real-world chart scenarios

For template reference:

• references/templates.md - Inline template documentation

Example Usage

Example 1: Simple Web App

User: "Create a Helm chart for my Node.js API"

Claude: I'll create a chart for your Node.js API. A few quick questions:
1. What's the chart name? (e.g., "user-api")
2. What's your container image? (e.g., "myorg/user-api:1.0.0")
3. Do you need an Ingress? (yes/no)

User: Name is user-api, image is myorg/user-api:1.0.0, yes to ingress

Claude: [Executes]
python3 scripts/scaffold_chart.py user-api \
  --workload-type deployment \
  --output /mnt/user-data/outputs \
  --ingress

[Chart generated successfully]
[Provides testing instructions]

Example 2: Database

User: "Create a Helm chart for PostgreSQL with persistent storage"

Claude: I'll create a StatefulSet chart for PostgreSQL.
What storage size do you need? (e.g., 10Gi, 20Gi)

User: 20Gi

Claude: [Executes]
python3 scripts/scaffold_chart.py postgresql \
  --workload-type statefulset \
  --output /mnt/user-data/outputs

[Then manually adds persistence config to values.yaml]
persistence:
  enabled: true
  size: 20Gi
  storageClass: standard

Example 3: Scheduled Backup

User: "Create a CronJob chart for nightly backups"

Claude: [Executes]
python3 scripts/scaffold_chart.py db-backup \
  --workload-type cronjob \
  --output /mnt/user-data/outputs

[Then updates values.yaml with schedule]
cronjob:
  schedule: "0 2 * * *"

Troubleshooting

Issue: Script fails to find templates Solution: Ensure running from skill directory or provide full path to script

Issue: CHARTNAME not replaced Solution: Script automatically handles this; if manual, use find/replace

Issue: Generated chart fails lint Solution: Check values.yaml for required fields, review templates

Notes

• Always use scaffold_chart.py when possible - it's faster and consistent
• Templates use CHARTNAME placeholder - script replaces automatically
• For complex customization, read templates from assets and modify manually
• Multi-environment configs are created separately after initial generation
• Testing instructions are critical - always include them