name	azure-services
description	Use when integrating Microsoft Azure services in this repo, including Azure AI, Azure OpenAI, Azure AI Search, Blob Storage, Key Vault, App Configuration, Service Bus, or Managed Identity.

Azure Services Skill

Use this skill for any task that adds, changes, or troubleshoots Azure-backed features.

Start by identifying the Azure service, the data flow, and whether the code runs in the browser, on the server, or in an Azure-hosted runtime.
Prefer the official Azure SDK for the target service. Use raw REST only when there is no SDK fit.
Keep Azure work additive to the existing Supabase/OpenAI app unless the user asks for a migration.

Never expose Azure API keys, connection strings, or tenant secrets to the browser.
Prefer managed identity or workload identity in deployed environments.
Use environment variables for local development and document them in the repo when they change.
If the feature needs secrets at runtime, assume they belong in server code or Key Vault, not in client code.

For Azure AI or Azure OpenAI features, keep calls server-side and return only the minimum response needed by the UI.
For search, document retrieval, or file processing flows, keep indexing, storage, and inference boundaries explicit.
Add guardrails for rate limits, safety behavior, and retry handling where relevant.

Validate configuration lazily inside the function that needs it unless the module is intentionally server-only.
Update any env example or setup documentation when a new Azure setting is introduced.
Prefer a concrete end-to-end verification step over assumptions when the integration path is new.

name	azure-services
description	Use when integrating Microsoft Azure services in this repo, including Azure AI, Azure OpenAI, Azure AI Search, Blob Storage, Key Vault, App Configuration, Service Bus, or Managed Identity.

Use this skill for any task that adds, changes, or troubleshoots Azure-backed features.

Start by identifying the Azure service, the data flow, and whether the code runs in the browser, on the server, or in an Azure-hosted runtime.
Prefer the official Azure SDK for the target service. Use raw REST only when there is no SDK fit.
Keep Azure work additive to the existing Supabase/OpenAI app unless the user asks for a migration.

Never expose Azure API keys, connection strings, or tenant secrets to the browser.
Prefer managed identity or workload identity in deployed environments.
Use environment variables for local development and document them in the repo when they change.
If the feature needs secrets at runtime, assume they belong in server code or Key Vault, not in client code.

For Azure AI or Azure OpenAI features, keep calls server-side and return only the minimum response needed by the UI.
For search, document retrieval, or file processing flows, keep indexing, storage, and inference boundaries explicit.
Add guardrails for rate limits, safety behavior, and retry handling where relevant.

Validate configuration lazily inside the function that needs it unless the module is intentionally server-only.
Update any env example or setup documentation when a new Azure setting is introduced.
Prefer a concrete end-to-end verification step over assumptions when the integration path is new.

azure-services