Migrate AI Gateway infrastructure to standalone RP and aliases

.github/workflows/ci.yml

@@ -0,0 +1,49 @@
+name: CI
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out repository
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.13"
+
+      - name: Set up uv
+        uses: astral-sh/setup-uv@v6
+        with:
+          enable-cache: true
+
+      - name: Install Bicep
+        run: az bicep install
+
+      - name: Build Bicep templates
+        run: |
+          az bicep build --file infra/main.bicep
+          az bicep build --file infra/ai-gateway/main.bicep
+
+      - name: Run infrastructure tests
+        run: |
+          bash tests/test-apim-lifecycle.sh
+          bash tests/test-ai-gateway-model-registration.sh
+
+      - name: Run Python unit tests
+        run: uv run --frozen python -m unittest discover -s tests -v
+
+      - name: Build container image
+        run: docker build --tag simple-foundry-hosted-agent-python-aigateway:ci .
+
+      - name: Run container smoke test
+        run: python tests/container_smoke.py --image simple-foundry-hosted-agent-python-aigateway:ci

IMPLEMENTATION_NOTES.md

@@ -263,31 +263,33 @@ The root deployment creates three resource groups:
 - `foundryagents` for the Foundry project, agent hosting, storage, Container
   Registry, and hosted-agent monitoring.
 - `foundrymodels` for the separate Foundry model account and deployments.
-- `gateway` for AI Gateway, Connector Namespace, Gateway monitoring, model
-  provider, and model registrations.
+- `gateway` for the standalone AI Gateway, Connector Namespace, Gateway
+  monitoring, model provider, model registrations, and model aliases.
 
 The only cross-group runtime authorization is the AI Gateway system identity's
 Foundry User role assignment on the model account. The public Azure
 role-definition ID used by the template is
 `53ca6127-db72-4b80-b1b0-d745d6d5456d`.
 
-The AI Gateway module uses these preview contracts:
+The AI Gateway module uses the live `2025-09-01-preview` contracts:
 
-- `Microsoft.ApiManagement/service@2025-09-01-preview`
-- SKU `AIGateway` with capacity `1`
-- `Microsoft.ApiManagement/service/workspaces/modelProviders@2025-09-01-preview`
-- `Microsoft.ApiManagement/service/workspaces/modelProviders/models@2025-09-01-preview`
-- `Microsoft.ApiManagement/service/workspaces/toolServers@2025-09-01-preview`
-- `Microsoft.ApiManagement/service/apiKeys@2025-09-01-preview`
+- `Microsoft.ApiManagement/aigateways` with SKU `Consumption`
+- `Microsoft.ApiManagement/aigateways/workspaces`
+- `Microsoft.ApiManagement/aigateways/workspaces/modelProviders`
+- `Microsoft.ApiManagement/aigateways/workspaces/modelProviders/models`
+- `Microsoft.ApiManagement/aigateways/workspaces/aliases`
+- `Microsoft.ApiManagement/aigateways/workspaces/toolServers`
+- `Microsoft.ApiManagement/aigateways/workspaces/telemetryExporters`
+- `Microsoft.ApiManagement/aigateways/apiKeys`
 - `Microsoft.Web/connectorGateways@2026-05-01-preview`
 
 The Connector Namespace uses the same name, resource group, and location as
-the APIM AI Gateway service. It is a separate resource from the GitHub MCP
+the AI Gateway. It is a separate resource from the GitHub MCP
 ToolServer and supports connector-backed scenarios.
 
-The default workspace is implicit for control-plane child resources. The
-postprovision hook waits for the workspace child collection before configuring
-the GitHub ToolServer.
+The predefined `workspaces/default` resource is the parent of every
+workspace-scoped child. The postprovision hook waits for the workspace child
+collection before configuring the GitHub ToolServer.
 
 The Foundry model provider uses:
 
@@ -302,12 +304,17 @@ Each model registration includes:
 - `apiFormat: OpenAIChatCompletions`
 - `/openai/v1/chat/completions`
 - `/openai/v1/responses`
-- A token-limit policy derived from deployment capacity
+- The live `displayName` wire property
 - `deployment.modelName` set to the Foundry deployment name from the final
   segment of the deployment resource ID
 
-The deployment name, not the backing catalog model name, is the OpenAI `model`
-value sent through AI Gateway.
+Model creation omits inline policies for compatibility with the live RP. The
+alias `gpt-latest` routes to the full deployment at priority 0 and fails over to
+the mini deployment at priority 1. Both backends explicitly set a weight of 100;
+aliases can instead place multiple backends at the same priority with relative
+weights for traffic splitting. The separate `gpt-mini-latest` alias addresses
+the mini deployment directly. Each alias backend points to the gateway child
+model resource ID, while the client sends the alias name as the OpenAI `model`.
 
 ## Regions and model defaults
 
@@ -336,19 +343,18 @@ The AI Gateway preview region allowlist is:
 
 The model defaults are:
 
-| Purpose | Gateway deployment | Backing model | Version | Capacity |
+| Purpose | Foundry deployment / Gateway alias | Backing model | Version | Capacity |
 | --- | --- | --- | --- | --- |
 | Full | `gpt-latest` | `gpt-5.6-sol` | `2026-07-09` | 20 |
 | Mini | `gpt-mini-latest` | `gpt-5.4-mini` | `2026-03-17` | 200 |
 
 If `gpt-5.6-sol` is unavailable, set `modelName=gpt-5.5` and
 `modelVersion=2026-04-24`.
 
-Global Standard capacity maps to the Gateway token limit at 1,000 tokens per
-minute per capacity unit. Capacity 20 therefore registers 20,000 tokens per
-minute for the full deployment. Mini capacity 200 accommodates requests from
-the GitHub Copilot app, also called ghapp, when built-in tool definitions make
-the prompt substantially larger.
+Global Standard deployment capacity is provisioned on the Foundry model
+account. Mini capacity 200 accommodates requests from the GitHub Copilot app,
+also called ghapp, when built-in tool definitions make the prompt substantially
+larger.
 
 The agent caps each response at 2,048 output tokens. Azure quota evaluation
 uses an estimate based on the prompt and requested maximum output, so this cap
@@ -371,17 +377,17 @@ Foundry fallback.
 
 ## AI Gateway deletion lifecycle
 
-APIM deletion can include a live service phase, a soft-deleted service phase,
-and an identity cleanup interval. Name availability alone does not prove that
-identity cleanup has completed.
+AI Gateway deletion has a live resource phase and an identity cleanup interval.
+The standalone `aigateways` resource does not use the APIM `service` deleted
+services endpoint. Name availability alone does not prove that identity cleanup
+has completed.
 
 `azure.yaml` invokes the lifecycle scripts in two paths:
 
 - Preprovision preserves a succeeded Gateway and removes only a terminal
   failed Gateway that matches the current azd environment ownership markers.
 - Postdown completes the explicit teardown, waits for the live resource to
-  disappear, purges the soft-deleted APIM service, and waits for identity
-  cleanup.
+  disappear, and waits for identity cleanup.
 
 The scripts use bounded exponential polling. Defaults are:
 
@@ -401,12 +407,6 @@ The nonsecret `.azure/<environment>/apim-lifecycle.state` marker contains only
 environment and resource identity, location, lifecycle state, time, and a
 deployment or resource failure identifier.
 
-Relevant public documentation:
-
-- [Azure API Management soft-delete](https://learn.microsoft.com/azure/api-management/soft-delete)
-- [Deleted Services - Get By Name](https://learn.microsoft.com/rest/api/apimanagement/deleted-services/get-by-name?view=rest-apimanagement-2024-05-01)
-- [Deleted Services - Purge](https://learn.microsoft.com/rest/api/apimanagement/deleted-services/purge?view=rest-apimanagement-2024-05-01)
-
 ## Hosted-agent image and monitoring
 
 The hosted agent uses `language: docker` and `docker.remoteBuild: true`. The

README.md

@@ -69,14 +69,31 @@ Or use the console:
 uv run chat.py
 ```
 
+Run the zero-auth container smoke test after building the image:
+
+```bash
+docker build -t daily-repo-digest .
+python tests/container_smoke.py --image daily-repo-digest
+```
+
+With an existing Azure login and selected azd environment, verify the deployed
+agent without redeploying, or optionally deploy first:
+
+```bash
+bash tests/test-azd-smoke.sh
+bash tests/test-azd-smoke.sh --deploy
+# Or provision and deploy first:
+bash tests/test-azd-smoke.sh --up
+```
+
 ## Deploy
 
 ```bash
 azd up
 azd ai agent invoke "Create a concise daily repo digest for microsoft/agent-framework."
 ```
 
-If Azure leaves a failed AI Gateway after an activation or managed-identity conflict, rerun `azd up`. The preprovision hook deletes only the terminal-Failed `AIGateway` tagged for the current azd environment, purges its APIM soft-delete record, and waits for identity cleanup. `azd down` performs the same bounded purge and settle process in its postdown hook. If the bound expires, follow the exact retry guidance printed by the hook rather than relying on name availability alone.
+If Azure leaves a failed AI Gateway after an activation or managed-identity conflict, rerun `azd up`. The preprovision hook deletes only the terminal-Failed `Microsoft.ApiManagement/aigateways` resource tagged for the current azd environment and waits for identity cleanup. `azd down` performs the same bounded settle process in its postdown hook. If the bound expires, follow the exact retry guidance printed by the hook rather than relying on name availability alone.
 
 Run the scheduled digest immediately:
 

github_mcp_middleware.py

@@ -86,10 +86,11 @@ def _normalize_github_arguments(
         workflow_filter = arguments.get("workflow_runs_filter")
         normalized_filter = dict(workflow_filter) if isinstance(workflow_filter, Mapping) else {}
         normalized_filter["status"] = "completed"
+        arguments.pop("per_page", None)
         arguments.update(
             method="list_workflow_runs",
             workflow_runs_filter=normalized_filter,
-            per_page=100,
+            perPage=100,
             page=1,
         )
 

infra/ai-gateway/main.bicep

@@ -8,8 +8,6 @@ param enableMonitoring bool = false
 param deployApimGatewayViaBicep bool = true
 param appInsightsName string = '${gatewayName}-insights'
 param logAnalyticsWorkspaceName string = take('${gatewayName}-logs', 63)
-param publisherEmail string = 'noreply@example.com'
-param publisherName string = 'AI Gateway'
 param deployFoundryProvider bool = false
 param foundryProviderName string = 'foundry'
 param foundryProviderDisplayName string = 'Foundry'
@@ -18,6 +16,7 @@ param foundryEndpoint string = ''
 param foundryResourceIds array = []
 param foundryManagedIdentityResource string = 'https://cognitiveservices.azure.com/'
 param foundryModels array = []
+param modelAliasesConfig array = []
 @secure()
 param toolServerConfigs object = {}
 
@@ -28,21 +27,18 @@ var logAnalyticsApiVersion = '2023-09-01'
 var telemetryWorkspaceName = 'default'
 var telemetryExporterName = 'appinsights'
 var normalizedFoundryEndpoint = endsWith(foundryEndpoint, '/') ? foundryEndpoint : '${foundryEndpoint}/'
-resource aiGateway 'Microsoft.ApiManagement/service@2025-09-01-preview' = if (deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource aiGateway 'Microsoft.ApiManagement/aigateways@2025-09-01-preview' = if (deployApimGatewayViaBicep) {
   name: gatewayName
   location: gatewayLocation
   tags: tags
   identity: {
     type: 'SystemAssigned'
   }
   sku: {
-    name: 'AIGateway'
-    capacity: 1
-  }
-  properties: {
-    publisherEmail: publisherEmail
-    publisherName: publisherName
+    name: 'Consumption'
   }
+  properties: {}
 }
 
 resource connectorNamespace 'Microsoft.Web/connectorGateways@2026-05-01-preview' = if (deployApimGatewayViaBicep) {
@@ -54,7 +50,8 @@ resource connectorNamespace 'Microsoft.Web/connectorGateways@2026-05-01-preview'
   ]
 }
 
-resource defaultWorkspace 'Microsoft.ApiManagement/service/workspaces@2025-09-01-preview' existing = if (deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource defaultWorkspace 'Microsoft.ApiManagement/aigateways/workspaces@2025-09-01-preview' existing = if (deployApimGatewayViaBicep) {
   parent: aiGateway
   name: telemetryWorkspaceName
 }
@@ -95,7 +92,8 @@ resource appInsights 'Microsoft.Insights/components@2020-02-02' = if (enableMoni
   }
 }
 
-resource telemetryExporter 'Microsoft.ApiManagement/service/workspaces/telemetryExporters@2025-09-01-preview' = if (enableMonitoring && deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource telemetryExporter 'Microsoft.ApiManagement/aigateways/workspaces/telemetryExporters@2025-09-01-preview' = if (enableMonitoring && deployApimGatewayViaBicep) {
   parent: defaultWorkspace
   name: telemetryExporterName
   properties: {
@@ -108,7 +106,8 @@ resource telemetryExporter 'Microsoft.ApiManagement/service/workspaces/telemetry
   }
 }
 
-resource foundryProvider 'Microsoft.ApiManagement/service/workspaces/modelProviders@2025-09-01-preview' = if (deployFoundryProvider && deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource foundryProvider 'Microsoft.ApiManagement/aigateways/workspaces/modelProviders@2025-09-01-preview' = if (deployFoundryProvider && deployApimGatewayViaBicep) {
   parent: defaultWorkspace
   name: foundryProviderName
   dependsOn: [
@@ -131,7 +130,8 @@ resource foundryProvider 'Microsoft.ApiManagement/service/workspaces/modelProvid
   }
 }
 
-resource foundryCatalogModels 'Microsoft.ApiManagement/service/workspaces/modelProviders/models@2025-09-01-preview' = [for model in foundryModels: if (deployFoundryProvider && deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource foundryCatalogModels 'Microsoft.ApiManagement/aigateways/workspaces/modelProviders/models@2025-09-01-preview' = [for model in foundryModels: if (deployFoundryProvider && deployApimGatewayViaBicep) {
   parent: foundryProvider
   name: model.armName
   properties: {
@@ -144,42 +144,64 @@ resource foundryCatalogModels 'Microsoft.ApiManagement/service/workspaces/modelP
       modelName: last(split(model.resourceId, '/'))
       modelVersion: model.?modelVersion ?? null
     }
-    policies: model.?policies ?? [
-      {
-        type: 'tokenLimit'
-        period: 'minute'
-        count: model.?tokenLimit ?? 10000
-        counterKey: 'Identity'
-      }
-    ]
   }
 }]
 
-resource toolServers 'Microsoft.ApiManagement/service/workspaces/toolServers@2025-09-01-preview' = [for config in items(toolServerConfigs): if (deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource modelAliases 'Microsoft.ApiManagement/aigateways/workspaces/aliases@2025-09-01-preview' = [for alias in modelAliasesConfig: if (deployFoundryProvider && deployApimGatewayViaBicep) {
+  parent: defaultWorkspace
+  name: alias.armName
+  dependsOn: [
+    foundryCatalogModels
+  ]
+  properties: {
+    aliasName: alias.aliasName
+    description: alias.?description ?? null
+    backendModels: [for target in alias.backendModels: {
+      resourceId: resourceId(
+        'Microsoft.ApiManagement/aigateways/workspaces/modelProviders/models',
+        gatewayName,
+        telemetryWorkspaceName,
+        foundryProviderName,
+        target.modelArmName
+      )
+      priority: target.priority
+      weight: target.weight
+    }]
+  }
+}]
+
+#disable-next-line BCP081
+resource toolServers 'Microsoft.ApiManagement/aigateways/workspaces/toolServers@2025-09-01-preview' = [for config in items(toolServerConfigs): if (deployApimGatewayViaBicep) {
   parent: defaultWorkspace
   name: config.key
   properties: config.value
 }]
 
-resource runtimeApiKey 'Microsoft.ApiManagement/service/apiKeys@2025-09-01-preview' = if (deployApimGatewayViaBicep) {
+#disable-next-line BCP081
+resource runtimeApiKey 'Microsoft.ApiManagement/aigateways/apiKeys@2025-09-01-preview' = if (deployApimGatewayViaBicep) {
   parent: aiGateway
   name: 'default'
   properties: {
     displayName: 'Default hosted agent key'
   }
 }
 
-output apimGatewayId string = resourceId('Microsoft.ApiManagement/service', gatewayName)
+var modelAliasResourceIds = [for alias in modelAliasesConfig: resourceId('Microsoft.ApiManagement/aigateways/workspaces/aliases', gatewayName, telemetryWorkspaceName, alias.armName)]
+
+output apimGatewayId string = resourceId('Microsoft.ApiManagement/aigateways', gatewayName)
 output connectorNamespaceId string = deployApimGatewayViaBicep ? connectorNamespace.id : ''
 output appInsightsId string = enableMonitoring ? appInsights.id : ''
 output appInsightsName string = enableMonitoring ? appInsights.name : ''
 output logAnalyticsWorkspaceId string = enableMonitoring ? logAnalyticsWorkspace.id : ''
 output logAnalyticsWorkspaceName string = enableMonitoring ? logAnalyticsWorkspace.name : ''
 output telemetryExporterId string = enableMonitoring && deployApimGatewayViaBicep ? telemetryExporter.id : ''
-output runtimeHostname string = deployApimGatewayViaBicep ? replace(aiGateway!.properties.gatewayUrl, 'https://', '') : ''
+output runtimeHostname string = deployApimGatewayViaBicep ? (aiGateway!.properties.?frontend.?defaultHostname ?? replace(aiGateway!.properties.?gatewayUrl ?? '', 'https://', '')) : ''
 output runtimeApiKeyId string = deployApimGatewayViaBicep ? runtimeApiKey.id : ''
 output gatewayPrincipalId string = deployApimGatewayViaBicep ? aiGateway!.identity.principalId : ''
 output foundryUserRoleAssignmentName string = deployFoundryProvider && deployApimGatewayViaBicep && !empty(foundryResourceIds) ? foundryUserRoleAssignments[0]!.outputs.roleAssignmentName : ''
+output modelAliasIds array = deployFoundryProvider && deployApimGatewayViaBicep ? modelAliasResourceIds : []
+output defaultModelAliasName string = !empty(modelAliasesConfig) ? modelAliasesConfig[0].aliasName : ''
 output gatewayLocation string = gatewayLocation
 output apiVersions object = {
   aiGateway: aiGatewayApiVersion