azure-review
Review an Azure solution architecture or implementation for platform fit, reliability, security, cost, operations, performance, governance, and deployment evidence. Use for Azure design reviews and Az
SKILL.mdskills/azure-review/SKILL.md
GitHubAzure Review
Assess the Azure design against documented requirements, current authoritative Azure guidance, and observable configuration. Do not infer subscriptions, regions, quotas, service features, policy assignments, or deployed state.
Workflow
- Read
AGENTS.md,architecture/toolkit-profile.yamlwhen present, requirements, ADD/C4/ADRs, infrastructure-as-code, application configuration, operational evidence, and the requested environments. - Inventory Azure services, regions, resource boundaries, identities, network/data flows, dependencies, scaling assumptions, availability targets, recovery targets, data residency, and cost constraints.
- Confirm time-sensitive service capabilities, limits, regional availability, retirement notices, and recommended patterns from current official Microsoft documentation. Cite the pages and access date in a persistent review.
- Evaluate the solution using references/azure-review-method.md, tailoring depth to the workload and applicable project profile.
- Validate important claims against IaC, policy/configuration, tests, monitoring, or runtime evidence. Label unobserved production claims
Not verified. - Separate architecture findings, deployment/configuration drift, operational-readiness gaps, and optimization opportunities. Give concrete remediation and verification steps.
- Persist only when requested, using the configured review path or
architecture/reviews/AZR-NNN-<slug>.md.
Rules
- Use Azure Well-Architected guidance as a review framework, not a substitute for workload requirements or evidence.
- Treat portal screenshots, diagrams, IaC, Azure Resource Graph results, policy state, and runtime telemetry according to what each actually proves.
- Check tenant/subscription/resource-group boundaries, identity and RBAC, networking, private access, encryption/key ownership, diagnostics, backup/restore, zone/region failure, scaling, quotas, deployment safety, policy, tagging, and cost controls when applicable.
- Identify service preview status, lock-in, regional constraints, unsupported combinations, and retirement/migration risk.
- Do not change Azure resources, policies, permissions, or deployments unless separately requested and authorized.
- Never claim Microsoft validation, compliance certification, or release approval.
Output
Lead with the Azure readiness result and blockers. Include scope and evidence freshness, workload inventory, pillar scorecard, findings by severity, resilience and recovery assessment, security/governance observations, cost and operability risks, unverified assumptions, and recommended owner/action.