Azure Scout: az-scout

Scout Azure regions for VM availability, zone mappings, pricing, spot scores, and quota — then plan deployments with confidence.

📖 Full documentation: docs.az-scout.com

📦 Plugin catalog: docs.az-scout.com/catalog

az-scout helps Azure operators and architects answer the questions that matter when planning resilient, cost-efficient VM deployments:

• Do my subscriptions share the same physical datacenter for logical zone 1?
• Which VM SKUs are available in all three zones with headroom in my quota?
• What is the Spot placement likelihood for this SKU family right now?
• Which deployment plan gives me the best confidence score across zones?

All from a single web UI — or through an MCP-powered AI agent connected to your favourite tools (Claude, VS Code Copilot, etc.).

Quick start

Prerequisites

Run locally with uv tool (recommended)

# Make sure you are authenticated to Azure
az login

# Run the tool (no install required)
uvx az-scout

Your browser opens automatically at http://127.0.0.1:5001.

Plugins

az-scout can be extended with pip-installable plugins discovered automatically at startup. See docs/PLUGINS.md for the plugin development guide and the scaffold for a ready-to-use template.

A Plugin Manager is included in the UI to browse the plugin catalog, install, update, and remove plugins.

Installation options

Recommended: install az-scout with uv

uv install az-scout
uvx az-scout

Alternative: install az-scout with pip

pip install az-scout
az-scout

Docker

docker run --rm -p 8000:8000 \
  -e AZURE_TENANT_ID=<your-tenant> \
  -e AZURE_CLIENT_ID=<your-sp-client-id> \
  -e AZURE_CLIENT_SECRET=<your-sp-secret> \
  ghcr.io/az-scout/az-scout:latest

Dev Container

A Dev Container configuration is included for a one-click development environment. Requires Docker and VS Code with the Dev Containers extension.

Open the repo in VS Code → Reopen in Container → run uv run az-scout web --host 0.0.0.0 --port 5001 --reload --no-open -v.

Azure Container App

It is also possible to deploy az-scout as a web app in Azure using the provided Bicep template (see Deploy to Azure section below).

Note: The web UI is designed for local use and may not be suitable for public-facing deployment without additional security measures (authentication, network restrictions, etc.). The MCP server can be exposed over the public internet if needed, but ensure you have proper authentication and authorization in place to protect access to Azure data.

UI guided deployment

A Bicep template is provided to deploy az-scout as an Azure Container App with a managed identity. You can use the Deploy to Azure button above for a portal-guided experience, or use the CLI commands below.

Resources created

The deployment creates:

Note: The Virtual Machine Contributor role is required for querying Spot Placement Scores (POST endpoint). Set enableSpotScoreRole=false to skip this if you don't need spot scores or prefer to manage permissions manually.

Usage

CLI options

az-scout [COMMAND] [OPTIONS]

az-scout --help.     # show global help
az-scout web --help  # show web subcommand help
az-scout mcp --help  # show mcp subcommand help
az-scout chat --help # show chat subcommand help
az-scout --version   # show version

az-scout web (default)

Run the web UI. This is the default when no subcommand is given.

  --host TEXT     Host to bind to.  [default: 127.0.0.1]
  --port INTEGER  Port to listen on.  [default: 5001]
  --no-open       Don't open the browser automatically.
  -v, --verbose   Enable verbose logging.
  --reload        Auto-reload on code changes (development only).
  --help          Show this message and exit.

az-scout mcp

Run the MCP server.

  --http          Use Streamable HTTP transport instead of stdio.
  --port INTEGER  Port for Streamable HTTP transport.  [default: 8080]
  -v, --verbose   Enable verbose logging.
  --help          Show this message and exit.

az-scout chat

Interactive AI chat in the terminal. Requires Azure OpenAI credentials.

# Start an interactive session
az-scout chat

# One-shot query
az-scout chat "What SKUs are available in francecentral?"

  -v, --verbose   Enable verbose logging.
  --help          Show this message and exit.

Slash commands are available during the chat session:

MCP server

An MCP server is included, allowing AI agents (Claude Desktop, VS Code Copilot, etc.) to query zone mappings and SKU availability directly.

Available tools

Plugin tools: Plugins can register additional MCP tools. For example, the Strategy Advisor plugin adds a capacity_strategy tool.

API

API documentation is available at /docs (Swagger UI) and /redoc (ReDoc) when the server is running.

License

MIT

Disclaimer

This tool is not affiliated with Microsoft. All capacity, pricing, and latency information are indicative and not a guarantee of deployment success. Spot placement scores are probabilistic. Quota values and pricing are dynamic and may change between planning and actual deployment. Latency values are based on Microsoft published statistics and must be validated with in-tenant measurements.