azure-pipelines

Best Practices

• Use YAML pipelines over classic UI pipelines for version-controlled, reviewable CI/CD
• Define trigger and pr sections with branch filters and path exclusions for targeted execution
• Use template references (extends, template) for reusable pipeline components across repos
• Leverage Azure-hosted agents with vmImage (ubuntu-latest, windows-latest, macos-latest)
• Use variable groups linked to Azure Key Vault for centralized secret management
• Define stages for multi-environment deployments with dependsOn for ordering
• Use deployment jobs with environment keyword for approval gates and deployment history
• Add condition expressions for conditional stage/job/step execution (succeeded(), failed(), eq())
• Use task versioning (e.g., NodeTool@0, Docker@2) to ensure reproducible builds
• Include displayName for all stages, jobs, and steps for readable pipeline runs
• Use caching with Cache@2 task for dependency directories (node_modules, .pip, .m2)
• Define parameters for configurable template inputs with type, default, and values
• Use strategy.matrix for multi-configuration testing (OS, language version combinations)
• Publish and download artifacts between stages using publish/download steps

Documentation Sources