Skip to content

Latest commit

 

History

17 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ACA Sandboxes

Secure, isolated compute environments with sub-second startup for agentic workloads.

Azure Container Apps Sandboxes is generally available.

Quick start

Prerequisites

  • Azure subscription with permission to create resource groups
  • Azure CLI (az) installed and logged in
  • uv (installs Python 3.13+ automatically)

Install and provision

uv sync
uv run python/samples/setup/setup.py

This creates a resource group, a sandbox group, grants your user the data-plane role, and writes python/samples/.env. Role assignments take 30-60 seconds to propagate.

Optionally install the aca CLI for shell-based samples.

Run your first sample

cd python/samples/01-webapps/simple-anonymous/python
uv run run.py

A sandbox boots, a Node.js app starts, and a public URL is printed. Open it in a browser to see a live system-stats page served from inside the sandbox. From here, pick any sample below.

Create and manage sandboxes

Method Description Link
Portal Browser UI for sandbox groups and sandboxes sandboxes.azure.com
CLI Shell scripting and automation CLI quickstart
Python SDK Programmatic access PyPI

Samples

Labs (Jupyter notebooks, Python SDK)

# Lab What it shows
01 getting-started Full surface end-to-end: create group → sandbox → exec → files → ports → egress → lifecycle → cleanup
02 bring-your-own-container Sandbox from your own container image, open a port to access its web content
03 sandbox-inception SDK inside a sandbox spawning child sandboxes via managed identity

Scenarios (composed use cases)

Each scenario includes a README with architecture notes, prerequisites, and step-by-step instructions.

# Scenario What it shows Python CLI
01 webapps Web app in a sandbox, anonymous or Entra-gated Python CLI
02 coding-agents Copilot CLI in a sandbox with deny-default egress Python CLI
03 code-interpreter LLM-driven code execution: generate, run, observe, iterate Python -
04 swarms Orchestrator coordinating many sandbox workers Python CLI
05 data-processing Producer/consumer pipelines on shared AzureBlob volumes Python -
06 developer-workflows PR builds, ephemeral CI, on-demand dev environments Python -
07 computer-use LLM computer-use agent driving Chrome inside a sandbox - -
08 sandbox-agents Agent frameworks (OpenAI, Claude, LangChain) using sandboxes as tool-execution backend Python -
09 mcp-hosting Host MCP servers in a sandbox Python -
10 connectors-email-triage Outlook trigger → sandbox → Copilot CLI → Teams triage card Python -
11 connectors-document-automation SharePoint trigger → sandbox → document extraction → results uploaded Python -

More resources

About

Official samples for ACA Sandboxes https://sandboxes.azure.com/docs/sandboxes/

Resources

Code of conduct

Contributing

Stars

18 stars

Watchers

1 watching

Forks

Releases

Packages

Used by

Contributors

Languages