Azure Omni-Tool

You deploy the Azure Omni-Tool MCP Server to bridge natural language requests with Azure CLI actions. It provides a safe planning-execution flow, audit trails, retry logic, and caching, enabling you to query Azure resources, run commands, and retrieve results through an MCP client with clear planning and execution steps.

How to use

You interact with the MCP server through a client that speaks the MCP protocol. Describe your Azure task in natural language; the server evaluates whether to plan or execute, enforces safety checks, and then runs the corresponding Azure CLI commands. You get structured results, including any planned steps, command outputs, and error analyses if something goes wrong. All actions are tagged for traceability via the audit trail.

Typical usage patterns include: planning complex multi-step operations before execution, retrying transient failures automatically, and leveraging tenant scoping to operate within a specific Azure tenant and subscription. When you request a command, you can choose to have the server return a plan first or proceed directly to execution if you’re confident in the action.

How to install

# 1. Install dependencies
npm install

# 2. Configure environment
cp .env.example .env
# Edit .env with your settings

# 3. Build & run
npm run build
npm start

The MCP client configuration to connect to the server is shown below. This enables the client to load the Azure MCP server and send tool requests for planning and execution.

{
  "mcpServers": {
    "azure_omni_tool": {
      "command": "node",
      "args": ["path/to/Azure-mcp/dist/index.js"]
    }
  }
}

Environment and security notes

Configure the environment to scope access and enable auditability. Required variables include Azure tenant and subscription identifiers, an operator email for audit trails, and logging and cache settings. Review and set the following environment variables before starting the server to ensure proper operation and traceability:

AZURE_TENANT_ID=
AZURE_SUBSCRIPTION_ID=
OPERATOR_EMAIL=
OPERATOR_NAME=
LOG_LEVEL=info
ENABLE_CACHE=true
CACHE_TTL_SECONDS=300
CACHE_CLEANUP_INTERVAL_MS=60000
MAX_RETRIES=3
RETRY_DELAY_MS=1000
COMMAND_TIMEOUT_MS=120000
AZURE_MCP_INCLUDE_PRODUCTION_CREDENTIALS=false

Notes on architecture and flow

The server exposes a Plan/Execute flow with safety guardrails to block unsafe commands and provide warnings when potentially destructive actions are detected. An audit trail records operator details for traceability. A caching layer reduces repeated queries, and a retry mechanism handles transient Azure API or CLI failures. Azure service adapters provide type-safe access to multiple services, while the CLI executor runs az CLI commands against Azure.

Troubleshooting tips

If you encounter failures during execution, check the audit logs for the operator email and correlation IDs, review the retry counts, and inspect the CLI timeout settings. Ensure your .env file is correctly populated and that Azure credentials are valid. If a plan-only run returns unexpected results, iterate on the explanation and plan to refine the requested actions.

Notes

This MCP server supports planning and executing Azure CLI commands with safety checks and an audit trail. It caches results for faster responses and can operate within a defined tenant and subscription context. Service adapters cover storage, cosmos, search, kusto, monitor, appconfig, keyvault, and postgres resources.

Available tools

Plan/Execute tool

Plan and execute Azure CLI commands with safety checks and an audit trail.

get_azure_context

Query Azure environment with caching for subscriptions, resource groups, and resources.

azure_service

Interact with specific Azure services such as storage, cosmos, search, kusto, monitor, appconfig, keyvault, and postgres.