Server has 7 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-50 chars, below the 50-200 char baseline for LLM optimization) and input schemas with type definitions. However, parameter descriptions are minimal or missing context. No output schemas documented. Error handling is absent, tools return raw results without recovery guidance. The auth_context parameter (object type, no schema) is a security red flag. Tool names follow verb_noun convention (get_, list_, start_, stop_, restart_) but descriptions lack WHEN/WHY context needed for LLM selection.

• Remove auth_context parameter entirely. Inject Azure credentials server-side via environment variables or Azure SDK DefaultAzureCredential. Never expose auth in tool parameters.
• Document output schemas for all tools. Example: get_vm_status should return {vm_id, vm_name, status, resource_group, subscription_id, power_state, provisioning_state}. This enables LLM chaining.
• Expand tool descriptions to 80-150 characters. Include WHEN to use: 'Get the current power state and provisioning status of an Azure VM. Use this before start_vm or stop_vm to check if the operation is safe.'
• Add error handling with recovery guidance. Example: 'VM not found in resource group. Verify resource_group and subscription_id are correct. Call list_vms to see available VMs.'
• Add confirmation step for destructive operations. Implement a dry_run parameter (default true) or a separate confirm_vm_operation tool that returns a confirmation_token required by start_vm/stop_vm/restart_vm.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry