preset

Deploy Model to Optimal Region

Automates intelligent Azure OpenAI model deployment by checking capacity across regions and deploying to the best available option.

What This Skill Does

1. Verifies Azure authentication and project scope
2. Checks capacity in current project's region
3. If no capacity: analyzes all regions and shows available alternatives
4. Filters projects by selected region
5. Supports creating new projects if needed
6. Deploys model with GlobalStandard SKU
7. Monitors deployment progress

Prerequisites

• Azure CLI installed and configured
• Active Azure subscription with Cognitive Services read/create permissions
• Microsoft Foundry project resource ID (PROJECT_RESOURCE_ID env var or provided interactively)
  • Format: /subscriptions/{sub-id}/resourceGroups/{rg}/providers/Microsoft.CognitiveServices/accounts/{account}/projects/{project}
  • Found in: Microsoft Foundry portal → Project → Overview → Resource ID
• The resource ID must match the format above exactly. Show the parsed subscription, resource group, account and project to the user and get confirmation before continuing. Stop if any part is empty or contains spaces, quotes or shell characters.

Minimal Permissions

This skill needs only three things. It reads the bundled reference files. It runs the az commands shown in the workflow. It runs the bundled deployment name script after reading it. It keeps capacity data in memory, leaves skill and user files as they are, and runs no other shell commands or web requests.

Quick Workflow

Fast Path (Current Region Has Capacity)

1. Check authentication → 2. Get project → 3. Check current region capacity
→ 4. Show the deployment plan, get user approval, then deploy

Alternative Region Path (No Capacity)

1. Check authentication → 2. Get project → 3. Check current region (no capacity)
→ 4. Query all regions → 5. Show alternatives → 6. Select region + project
→ 7. Deploy

---

Deployment Phases

Approval gate: Before any account create or deployment create, show the user the subscription, resource group, region, model, version, SKU and capacity. Note that these resources are billable. Proceed only after the user explicitly approves. Cost can be checked with the separate cost estimation skill.

Input checks: Take model name, version and region only from values returned by the list-models and capacity queries. Accept only letters, digits, hyphens and periods in model name, version, region and deployment name. Keep every value double-quoted in commands, and stop if a value contains anything else.

Capacity: Show the computed capacity and its effect on remaining quota, and let the user adjust it before deploying.

Bundled scripts: Before running any bundled script, read it and confirm that it only lists deployments or builds a portal link. If it does anything else, or is missing, stop and choose the deployment name by hand using the numeric suffix rule.

For detailed step-by-step instructions, see workflow reference.

---

Error Handling

---

Advanced Usage

# Custom capacity (must not exceed the available capacity reported by the query; confirm with the user first)
az cognitiveservices account deployment create ... --sku-capacity <value>

# Check deployment status
az cognitiveservices account deployment show --name <acct> --resource-group <rg> --deployment-name <name> --query "{Status:properties.provisioningState}"

Notes

• SKU: GlobalStandard only — API Version: 2024-10-01 (GA stable)
• Data residency: GlobalStandard and alternate regions may process data outside the geography of the project. Ask the user about residency constraints before showing alternatives, offer only regions the user approves, and use customize when data residency matters.

---

Related Skills

• microsoft-foundry - Parent skill for Microsoft Foundry operations
• quota — For quota viewing, increase requests, and troubleshooting quota errors, defer to this skill
• azure-quick-review - Review Azure resources for compliance
• cost-estimation - Estimate costs through the separately installed azure-cost plugin
• azure-validate - Validate Azure infrastructure before deployment