Azure Deployment Preflight

Azure Deployment Preflight Validation

Validate ARM template deployments before execution by running what-if analysis, checking permissions, and generating a structured preflight report.

Adapted from github/awesome-copilot azure-deployment-preflight skill.

When to Use

• Before deploying infrastructure to Azure
• To preview what changes a deployment will make
• To verify permissions are sufficient for deployment
• As part of the template generation workflow (invoked by template generator)

Procedure

Step 1: Locate Template Files

Find the ARM template and parameters to validate:

# From deployment artifacts
DEPLOYMENT_ID="deploy-20260218-193500"
TEMPLATE=".azure/deployments/$DEPLOYMENT_ID/template.json"
PARAMETERS=".azure/deployments/$DEPLOYMENT_ID/parameters.json"

# Or from user-provided path
TEMPLATE="path/to/template.json"

Verify files exist before proceeding.

Step 1.5: Resource Availability Gate

Invoke /azure-resource-availability skill to validate all resources in the template before running what-if.

This catches issues that az deployment sub validate and what-if do NOT catch — such as:

• VM SKU restrictions per subscription/region
• Deprecated or LTS-only service versions (e.g., Kubernetes)
• API version incompatibilities (fields that don't exist in the chosen version)
• Quota limits that would be exceeded

# Parse template for resources to check
# For each resource: extract type, apiVersion, SKU, service version, region
# Pass to /azure-resource-availability

Gate behavior:

• ✅ PASSED → continue to Step 2
• ⚠️ WARNINGS → continue but include warnings in preflight report
• ❌ BLOCKED → STOP. Report blocking issues with alternatives. Do not proceed to what-if.

Save the availability report to .azure/deployments/$DEPLOYMENT_ID/availability-report.md.

Step 2: Validate Template Syntax

Check the ARM template is valid JSON and follows schema:

# Validate template syntax
az deployment sub validate \
  --location {location} \
  --template-file "$TEMPLATE" \
  --parameters @"$PARAMETERS" \
  --output json 2>&1

What to capture:

• Syntax errors with details
• Schema validation warnings
• Parameter validation issues

Step 3: Run What-If Analysis

Run what-if to preview resource changes without deploying:

# For subscription-level deployments (Git-Ape default)
az deployment sub what-if \
  --location {location} \
  --template-file "$TEMPLATE" \
  --parameters @"$PARAMETERS" \
  --result-format FullResourcePayloads \
  --output json 2>&1

Fallback: If permission errors occur, retry without RBAC validation:

az deployment sub what-if \
  --location {location} \
  --template-file "$TEMPLATE" \
  --parameters @"$PARAMETERS" \
  --result-format FullResourcePayloads \
  --no-prompt \
  --output json 2>&1

Note the fallback in the report.

Step 4: Parse What-If Results

Categorize resource changes from the what-if output:

For modified resources, capture the specific property changes:

Resource: Microsoft.Storage/storageAccounts/starnwkdhk
  ~ properties.minimumTlsVersion: "TLS1_0" → "TLS1_2"
  ~ tags.Environment: "staging" → "dev"

Step 5: Generate Preflight Report

Create a structured report saved to .azure/deployments/$DEPLOYMENT_ID/preflight-report.md:

# Preflight Validation Report

**Deployment ID:** {deployment-id}
**Validated:** {timestamp}
**Template:** {template-path}
**Scope:** Subscription-level

## Summary

| Check | Status |
|-------|--------|
| Template Syntax | ✅ Valid |
| Schema Compliance | ✅ Valid |
| What-If Analysis | ✅ Completed |
| Permission Check | ✅ Sufficient |

## What-If Results

### Resources to Create ({count})
| Resource Type | Name | Location |
|---------------|------|----------|
| Microsoft.Resources/resourceGroups | rg-xxx | eastus |
| Microsoft.Storage/storageAccounts | stxxx | eastus |

### Resources to Modify ({count})
| Resource | Property | Current | New |
|----------|----------|---------|-----|
| stxxx | minimumTlsVersion | TLS1_0 | TLS1_2 |

### Resources to Delete ({count})
{list or "None"}

### Resources Unchanged ({count})
{count resources with no changes}

## Issues Found

### Errors
{any blocking errors}

### Warnings
{any non-blocking warnings}

## Recommendations
- {actionable suggestions}

Step 6: Return Results

Return the preflight report to the calling agent. If any errors were found, recommend the user address them before deploying.

Save report to .azure/deployments/$DEPLOYMENT_ID/preflight-report.md.

Error Handling

Key principle: Continue validation even when errors occur. Capture all issues in the final report.

Integration

Called by template generator after generating ARM template:

Template Generator → /azure-deployment-preflight → Preflight Report → User Confirmation

Manual invocation:

User: /azure-deployment-preflight --deployment-id deploy-20260218-193500

Agent: Running preflight validation...
[Validates template, runs what-if, generates report]