09-deploy-ai-models-from-local

name	09.deploy-ai-models-from-local
description	Automates local Azure AI Foundry model deployment workflow for only sections 9.1, 9.2, 9.3, and 9.6 of LOCAL_DEVELOPMENT.md.
argument-hint	Use this skill to list available models, validate Azure prerequisites, deploy models via Terraform, and configure API Foundry env vars.
compatibility	["linux","macos","windows"]
license	MIT
user-invokable	true

Skill: Azure AI Foundry — Model Deployment (Scoped)

This skill follows the same wrapper-first automation pattern used in previous local setup skills.

Scope is intentionally limited to:

• 9.1 Available Models
• 9.2 Prerequisites
• 9.3 Deploy AI Models with Terraform
• 9.6 Configure the API to Use AI Models

No other section 9 subsections are implemented by this skill.

---

Prerequisites

• Azure CLI installed.
• Terraform installed.
• Access to Azure subscription/resource group with AI Services permissions.
• infra/params/dev.tfvars exists (copy from tfvars.example and fill in your values if not already created).

---

Automated Setup (Recommended)

Use wrapper entrypoints. These wrappers call internal scripts and keep usage consistent.

• Windows/PowerShell wrapper: setup-ai-models-from-local.ps1
• Linux/macOS bash wrapper: setup-ai-models-from-local.sh

Maintainer note:

• Entry points are wrapper scripts.
• Internal implementation scripts are:
  • Manage-AIModelsFromLocal.ps1
  • manage-ai-models-from-local.sh

PowerShell wrapper

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action list-models

bash wrapper

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh list-models

---

9.1 Available Models

List model definitions from infra/config/ai-models.json:

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action list-models

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh list-models

---

9.2 Prerequisites

Checks and enforces prerequisites:

• az exists
• authenticated session (az login required if missing)
• optional subscription selection
• provider registration checks:
  • Microsoft.CognitiveServices
  • Microsoft.AppService

Run checks

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action check-prereqs -SubscriptionId "<subscription-id>"

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh check-prereqs --subscription-id "<subscription-id>"

---

9.3 Deploy AI Models with Terraform

Runs Terraform workflow in infra/:

• terraform init
• terraform plan -var-file=<tfvars>
• terraform apply -var-file=<tfvars>

Deploy with confirmation prompt

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action deploy-terraform -TfvarsPath "infra/params/dev.tfvars"

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh deploy-terraform --tfvars-path "infra/params/dev.tfvars"

Deploy with auto-approve

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action deploy-terraform -TfvarsPath "infra/params/dev.tfvars" -AutoApprove

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh deploy-terraform --tfvars-path "infra/params/dev.tfvars" --auto-approve

---

9.6 Configure the API to Use AI Models

Updates app/backend/.env values:

• FOUNDRY_ENDPOINT
• FOUNDRY_API_KEY
• APIM_OPENAI_URL (optional)

You can provide values explicitly, or infer endpoint/key from Azure using resource group + AI Services account name.

Option A: Explicit values

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action configure-api-env `
  -FoundryEndpoint "https://ai-pronghorn-dev.services.ai.azure.com/" `
  -FoundryApiKey "<foundry-api-key>" `
  -ApimOpenAiUrl "https://apim-pronghorn-xxx.azure-api.net/openai"

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh configure-api-env \
  --foundry-endpoint "https://ai-pronghorn-dev.services.ai.azure.com/" \
  --foundry-api-key "<foundry-api-key>" \
  --apim-openai-url "https://apim-pronghorn-xxx.azure-api.net/openai"

Option B: Infer endpoint/key from Azure account

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action configure-api-env `
  -ResourceGroupName "pronghorn-blue" `
  -AIServicesName "ai-pronghorn-dev" `
  -ApimOpenAiUrl "https://apim-pronghorn-xxx.azure-api.net/openai"

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh configure-api-env \
  --resource-group "pronghorn-blue" \
  --ai-services-name "ai-pronghorn-dev" \
  --apim-openai-url "https://apim-pronghorn-xxx.azure-api.net/openai"

---

One-Command Scoped Flow

Run all supported subsections in order: 9.1 -> 9.2 -> 9.3 -> 9.6.

Set-Location .github/skills/09.deploy-ai-models-from-local/scripts
.\setup-ai-models-from-local.ps1 -Action all -SubscriptionId "<subscription-id>" -TfvarsPath "infra/params/dev.tfvars"

cd .github/skills/09.deploy-ai-models-from-local/scripts
bash ./setup-ai-models-from-local.sh all --subscription-id "<subscription-id>" --tfvars-path "infra/params/dev.tfvars"

---

Validation

After successful run:

• Terraform apply completes without errors.
• app/backend/.env contains non-empty FOUNDRY_ENDPOINT and FOUNDRY_API_KEY.
• API can read Foundry configuration on startup.

---

Related Resources

• LOCAL_DEVELOPMENT.md § 9.1
• LOCAL_DEVELOPMENT.md § 9.2
• LOCAL_DEVELOPMENT.md § 9.3
• LOCAL_DEVELOPMENT.md § 9.6