pipeline-setup

name: pipeline-setup description: Creates Azure DevOps CI/CD pipelines from the template YAML files. This skill creates pipelines for agent creation, testing, and deployment automation using the ready-to-use pipeline definitions.

Pipeline Setup for Azure AI Foundry

This skill handles creating Azure DevOps pipelines from the template YAML files in your repository.

When to use this skill

Use this skill when you need to:

• Create CI/CD pipelines from template YAML files
• Set up automated agent deployment pipelines
• Configure pipeline triggers and branches
• Enable continuous integration and deployment

Prerequisites

Before using this skill, ensure:

• ✅ Configuration loaded (use configuration-management skill first)
• ✅ Repository created with template code (use repository-setup skill)
• ✅ Service connections configured (use service-connection-setup skill)
• ✅ Variable groups and environments created (use environment-setup skill)
• ✅ Azure DevOps authentication configured (bearer token set)

What This Skill Creates

• CI/CD Pipelines from template YAML files
• Pipeline definitions linked to your repository
• Automated triggers (optional) for continuous deployment

Automated Script

This skill includes scripts/create-pipelines.ps1 which automates the entire pipeline setup process:

What it does automatically:

1. ✅ Updates pipeline YAML files - Replaces REPLACE_WITH_YOUR_PROJECTNAME placeholder with actual config.naming.projectName
   • Variable groups: REPLACE_WITH_YOUR_PROJECTNAME-dev-vars → myproject-dev-vars
   • Service connections: REPLACE_WITH_YOUR_PROJECTNAME-dev → myproject-dev
2. ✅ Commits and pushes changes - Updates YAML files in the repository
3. ✅ Creates pipelines - Sets up pipelines from the updated YAML files

Usage:

cd .github/skills/pipeline-setup
./scripts/create-pipelines.ps1 -UseConfig

Why is this needed?

• Variable groups are named {projectName}-{env}-vars (e.g., myproject-dev-vars)
• Service connections are named {projectName}-{env} (e.g., myproject-dev)
• Pipeline YAML files must reference the correct names
• The script ensures consistency between created infrastructure and pipeline configuration

Pipeline Templates Available

The template application includes these pipeline YAML files:

Step-by-step execution

Step 1: Load Configuration

# Load configuration
. ./.github/skills/configuration-management/config-functions.ps1
$config = Get-StarterConfig

# Extract values
$org = $config.azureDevOps.organizationUrl
$project = $config.azureDevOps.projectName
$projectName = $config.naming.projectName
$repoName = "azure-ai-foundry-app"  # Your repository name

# Verify repository exists
$repoExists = az repos show --repository $repoName --query "id" -o tsv
if (-not $repoExists) {
    Write-Host "❌ Repository not found: $repoName"
    Write-Host "   Run the repository-setup skill first"
    exit 1
}

Write-Host "✓ Configuration loaded"
Write-Host "✓ Repository found: $repoName"
Write-Host "✓ Project name from config: $projectName"

Step 2: Update Pipeline YAML Files (Automated)

Important: The pipeline YAML files contain placeholders that need to be replaced with your actual project name.

The create-pipelines.ps1 script automatically:

1. Clones the repository temporarily
2. Replaces REPLACE_WITH_YOUR_PROJECTNAME with $projectName from config
3. Commits and pushes the updated YAML files back to the repository
4. Creates the pipelines from the updated YAML files

Example replacements:

# Variable groups - Before
variables:
  - group: 'REPLACE_WITH_YOUR_PROJECTNAME-dev-vars'

# Variable groups - After
variables:
  - group: 'myproject-dev-vars'  # Where myproject = config.naming.projectName

# Service connections - Before
- task: AzureCLI@2
  inputs:
    azureSubscription: 'REPLACE_WITH_YOUR_PROJECTNAME-dev'

# Service connections - After
- task: AzureCLI@2
  inputs:
    azureSubscription: 'myproject-dev'  # Where myproject = config.naming.projectName

Manual alternative (if needed):

# Clone repository
git clone https://dev.azure.com/$org/$project/_git/$repoName
cd $repoName

# Replace placeholders in YAML files (both variable groups and service connections)
$yamlFiles = Get-ChildItem -Path ".azure-pipelines" -Filter "*.yml"
foreach ($file in $yamlFiles) {
    (Get-Content $file.FullName) -replace "REPLACE_WITH_YOUR_PROJECTNAME", $projectName | Set-Content $file.FullName
}

# Commit and push
git add .azure-pipelines/*.yml
git commit -m "Update pipeline YAML files with projectName: $projectName"
git push origin main

Step 3: Create Pipelines from Templates

Write-Host "`nCreating pipelines..."

# Define pipelines to create
$pipelines = @(
    @{
        name = "Azure AI Foundry - Create Agent"
        path = ".azure-pipelines/createagentpipeline.yml"
        description = "Deploys AI agent to Azure AI Foundry"
    },
    @{
        name = "Azure AI Foundry - Agent Evaluation"
        path = ".azure-pipelines/agenteval.yml"
        description = "Runs agent evaluation tests"
    },
    @{
        name = "Azure AI Foundry - Red Team"
        path = ".azure-pipelines/redteam.yml"
        description = "Runs security red team tests"
    }
)

foreach ($pipeline in $pipelines) {
    # Check if pipeline already exists
    $existingPipeline = az pipelines list --query "[?name=='$($pipeline.name)'].id" --output tsv
    
    if (-not $existingPipeline) {
        Write-Host "Creating pipeline: $($pipeline.name)"
        
        try {
            $pipelineId = az pipelines create `
                --name "$($pipeline.name)" `
                --repository $repoName `
                --repository-type tfsgit `
                --branch main `
                --yml-path "$($pipeline.path)" `
                --skip-first-run `
                --output json | ConvertFrom-Json | Select-Object -ExpandProperty id
            
            Write-Host "✓ Pipeline created: $($pipeline.name) (ID: $pipelineId)"
            Write-Host "  Description: $($pipeline.description)"
            Write-Host "  YAML: $($pipeline.path)"
        }
        catch {
            Write-Host "❌ Failed to create pipeline: $($pipeline.name)"
            Write-Host "   Error: $_"
            Write-Host "   Verify YAML path exists in repository: $($pipeline.path)"
        }
    } else {
        Write-Host "✓ Pipeline already exists: $($pipeline.name) (ID: $existingPipeline)"
    }
}

Write-Host "`n✅ Pipeline setup complete!"

Step 4: Verify Pipeline Configuration

Write-Host "`n=== Verifying Pipeline Configuration ==="

# List all pipelines
$allPipelines = az pipelines list --output json | ConvertFrom-Json

Write-Host "`nCreated Pipelines:"
foreach ($p in $allPipelines) {
    Write-Host "  - $($p.name)"
    Write-Host "    ID: $($p.id)"
    Write-Host "    Path: $($p.path)"
    Write-Host "    Repository: $($p.repository.name)"
}

Write-Host "`nPipeline URLs:"
foreach ($p in $allPipelines) {
    Write-Host "  - $($p.name): $org/$project/_build?definitionId=$($p.id)"
}

Step 5: (Optional) Configure Pipeline Triggers

By default, pipelines are created with --skip-first-run to prevent automatic execution.

To enable CI triggers (automatic runs on code push):

# Option 1: Update YAML file to enable triggers
# Edit .azure-pipelines/createagentpipeline.yml:
# trigger:
#   branches:
#     include:
#       - main
#   paths:
#     include:
#       - src/agents/*

# Option 2: Enable via Azure DevOps UI
Write-Host "`nTo enable CI triggers:"
Write-Host "1. Go to: $org/$project/_build"
Write-Host "2. Select pipeline > Edit"
Write-Host "3. Click 'Triggers' tab"
Write-Host "4. Enable 'Continuous integration'"
Write-Host "5. Configure branch filters and path filters"

Step 6: (Optional) Run First Pipeline

Write-Host "`n=== Running First Pipeline (Optional) ==="

# Get the Create Agent pipeline ID
$createAgentPipeline = az pipelines list --query "[?name=='Azure AI Foundry - Create Agent'].id" --output tsv

if ($createAgentPipeline) {
    Write-Host "To run the Create Agent pipeline:"
    Write-Host "1. Manual run via CLI:"
    Write-Host "   az pipelines run --id $createAgentPipeline"
    Write-Host ""
    Write-Host "2. Manual run via UI:"
    Write-Host "   $org/$project/_build?definitionId=$createAgentPipeline"
    Write-Host "   Click 'Run pipeline' button"
    Write-Host ""
    Write-Host "3. Automatic run on code push (if CI triggers enabled)"
} else {
    Write-Host "⚠️  Create Agent pipeline not found"
}

Pipeline YAML Structure

Understanding the pipeline YAML structure helps with customization:

# Basic structure of template pipelines
trigger:
  branches:
    include:
      - main
  paths:
    include:
      - src/agents/*

variables:
  - group: REPLACE_WITH_YOUR_PROJECTNAME-dev-vars  # References variable group: {projectName}-dev-vars

stages:
  - stage: Dev
    jobs:
      - deployment: DeployAgent
        environment: dev  # References environment
        pool:
          vmImage: 'ubuntu-latest'
        strategy:
          runOnce:
            deploy:
              steps:
                - task: AzureCLI@2
                  inputs:
                    azureSubscription: 'azure-foundry-dev'  # References service connection

Troubleshooting

Common Issues and Solutions

1. Pipeline YAML Path Not Found

Error: Could not find file at path .azure-pipelines/createagentpipeline.yml

Solution: Verify the YAML file exists in your repository:

# Check files in repository
az repos list-branches --repository $repoName

# Or check locally
Set-Location "C:\Repos\ado\azure-ai-foundry-app"
Test-Path ".azure-pipelines/createagentpipeline.yml"

# If missing, ensure template code was pushed correctly (repository-setup skill)

2. Service Connection Not Found

Error: The pipeline is not valid. Could not find service connection 'azure-foundry-dev'

Solution: Verify service connections exist and are authorized:

# List service connections
az devops service-endpoint list --query "[].name" --output tsv

# Authorize service connection for all pipelines
$scId = az devops service-endpoint list --query "[?name=='azure-foundry-dev'].id" --output tsv
az devops service-endpoint update --id $scId --enable-for-all true

3. Variable Group Not Found

Error: The pipeline is not valid. Could not find variable group '{projectName}-dev-vars'

Solution: Verify variable groups exist and are authorized:

# List variable groups
az pipelines variable-group list --query "[].name" --output tsv

# Authorize variable group (use your actual projectName from config)
$vgName = "$projectName-dev-vars"  # Replace with your config.naming.projectName
$vgId = az pipelines variable-group list --query "[?name=='$vgName'].id" --output tsv
az pipelines variable-group update --id $vgId --authorize true

4. Environment Not Found

Error: The pipeline is not valid. Could not find environment 'dev'

Solution: Verify environments exist:

# List environments
az pipelines environment list --query "[].name" --output tsv

# Create if missing (use environment-setup skill)

5. Pipeline Creation Permission Denied

Error: TF401027: You need the Git 'Create Repository' permission

Solution: Verify you have necessary permissions:

1. Check Azure DevOps project permissions
2. Ensure you have "Build Administrator" role
3. Or use a PAT with "Build (read and execute)" and "Release (read, write, execute and manage)" scopes

6. Repository Type Error

Error: --repository-type must be github or tfsgit

Solution: Ensure you specify --repository-type tfsgit for Azure DevOps repositories:

az pipelines create `
    --repository-type tfsgit `  # Use tfsgit for Azure Repos
    ...

Best practices

1. Skip first run - Use --skip-first-run to prevent immediate execution
2. Review YAML before creating - Understand pipeline logic and dependencies
3. Use descriptive names - Follow pattern: "Azure AI Foundry - {Purpose}"
4. Enable CI triggers carefully - Start manual, enable automation later
5. Test one pipeline first - Verify Create Agent pipeline before others
6. Document customizations - Track any changes to template YAML files
7. Use pipeline templates - Create reusable YAML templates for common tasks
8. Monitor pipeline runs - Check run history and logs regularly

Pipeline Execution Order

Recommended order for first runs:

1. Create Agent - Deploy your first agent to AI Foundry
2. Agent Evaluation - Validate agent responses and behavior
3. Red Team - Test agent security and safety

Integration with other skills

This skill works together with:

• configuration-management (required first) - Loads centralized configuration
• repository-setup (required) - Creates repository with template YAML files
• service-connection-setup (required) - Pipelines reference service connections
• environment-setup (required) - Pipelines reference variable groups and environments
• deployment-validation (next step) - Validates pipeline execution

Related resources

• docs/troubleshooting.md - Lesson #2, #5: Pipeline issues
• docs/starter-guide.md - Complete deployment guide
• docs/azure-devops-cicd-reference.md - Pipeline reference
• Azure DevOps: Pipelines
• Azure DevOps: YAML schema