Azure AI Services

7 steps · start to finish.

1. 1
   Step 1

   Environment Setup

   ▶

   Verify the Azure environment and determine which invocation method to use (MCP vs. CLI).

   echo "=== AZURE AI SERVICES SETUP ==="
   
   # Check if Azure MCP is available
   if command -v mcp >/dev/null 2>&1; then
     echo "MCP available — will prefer azure__search / azure__speech tools"
     USE_MCP=true
   else
     echo "MCP not available — will use Azure CLI fallback"
     USE_MCP=false
   fi
   
   # Verify az CLI is installed (needed as fallback or for management tasks)
   command -v az >/dev/null 2>&1 || { echo "WARNING: az CLI not found. Install via: curl -sL https://aka.ms/InstallAzureCLIDeb | sudo bash"; }
   
   # Verify login state
   az account show --query "{subscription:id, name:name}" -o json 2>/dev/null || {
     echo "Not logged in to Azure. Run: az login"
     exit 1
   }
   
   # Create output directory
   mkdir -p /app/results
   
   echo "Setup complete."
   

   If the environment is not authenticated, abort and instruct the user to run az login or configure a service principal.

   ---

2. 2
   Step 2

   Select Service and Verify Resource Exists

   ▶

   Based on the `azure_service` parameter, confirm the target resource exists and is accessible.

3. 3
   Step 3

   Invoke the Azure AI Service

   ▶

   Choose the appropriate invocation path based on the selected service and available tools.

4. 4
   Step 4

   Iterate on Errors (max 3 rounds)

   ▶

   If Step 3 fails or returns unexpected results:

5. 5
   Step 5

   Validate Output

   ▶

   Confirm the service returned meaningful results before writing the summary.

6. 6
   Step 6

   Write Commands Log

   ▶

   Document all MCP tool calls and CLI commands executed during this run.

7. 7
   Step 7

   Final Checklist (MANDATORY — do not skip)

   ▶

   echo "=== FINAL OUTPUT VERIFICATION ===" RESULTS_DIR="/app/results" for f in \ "$RESULTS_DIR/summary.md" \