Persistent Linux machines
Each Sandbox runs as an isolated Linux machine with a persistent 50 GB disk. Files, installed packages, configuration, and caches remain available after restarts.
Buddy Sandboxes
Buddy Sandboxes provide isolated, persistent Linux machines for running AI-generated code, applications, tests, and development environments. They are designed for developers, teams, and agent-driven workflows that need retained files, command execution, application endpoints, and programmatic control.
Persistent cloud machines for code execution
Buddy Sandboxes are isolated Linux machines in Buddy’s cloud for running AI-generated code, applications, tests, and development environments. Each machine has allocated CPU and RAM, a persistent 50 GB disk, installed development tools, and public endpoints for exposing applications over HTTPS.
Unlike a throwaway execution container, a Sandbox keeps its disk state between sessions. Teams and agents can install packages, clone repositories, run commands, supervise applications, inspect logs, and connect over SSH. Sandboxes can be created and driven interactively from the CLI or automated through the SDK, REST API, YAML, pipelines, and MCP.
Key features
Root-level administration
The buddy user has passwordless sudo, allowing packages and system users to be installed and Docker to run inside the Sandbox.
Commands, files, and apps
Run commands with streamed output and exit codes, transfer files and directories, clone repositories or fetch artifacts, and manage long-running apps with separate logs.
Public application endpoints
Expose application ports through public HTTPS endpoints, with optional Buddy sign-in and request logging for inspecting incoming traffic.
Snapshots and repeatable environments
Use snapshots to create new Sandboxes from a prepared environment, including system packages, repositories, runtimes, and dependency caches.
Multiple management interfaces
Create and operate Sandboxes through the CLI, TypeScript SDK, REST API, YAML, pipelines, or MCP, depending on whether the work is interactive, automated, or agent-driven.
Use cases
AI coding workspaces
Give an agent an isolated Linux environment where it can install dependencies, edit files, run tests, and make changes without modifying a developer laptop or production system.
Preview environments
Create a machine for each branch or pull request, expose the running application through a public URL, and let teammates or clients review the result.
Automated testing
Run a full test environment, execute commands and application checks, collect exit codes and logs, then destroy the Sandbox when the test workflow is complete.
Application hosting
Keep an application running on a machine with no timeout unless one is configured, expose it through an endpoint, and use a custom domain through the supported Buddy workflow.
Remote development
Use SSH and VS Code Remote-SSH to work on a remote Linux environment with its files, packages, and configuration retained between sessions.
Pros and Cons
Pros
- Persistent local disk preserves files, packages, databases, and build caches between sessions.
- Passwordless sudo, SSH access, and Docker-in-Sandbox support enable conventional Linux development workflows.
- Commands, file transfers, applications, endpoints, logs, snapshots, and lifecycle operations are available through documented interfaces.
- Configured apps can restart on boot, while endpoint traffic or commands can wake a stopped Sandbox.
Cons
- Running processes and in-memory state do not survive a stop; only persistent disk data remains.
- The TypeScript SDK requires Node.js 20 or newer, is published before version 1.0, and may change between releases.
- The official SDK documentation lists TypeScript as the supported SDK language; other stacks use the REST API.
FAQ
What is Buddy Sandboxes?
A Sandbox is an isolated Linux machine with its own operating system, CPU and RAM allocation, persistent disk, applications, and public endpoints. It is designed to behave more like a persistent virtual machine than a disposable container.
What survives when a Sandbox restarts?
The persistent disk, including files, installed packages, database contents, and build caches, remains after a stop and restart. Memory does not persist: running processes terminate, while apps defined in the Sandbox configuration start again when the machine boots.
Do I need a Dockerfile?
No. You can configure a machine, save it as a snapshot, and create new Sandboxes from that prepared state. This avoids building and pushing a Docker image for that workflow.
How can Sandboxes be managed?
Sandboxes can be managed through the web UI, CLI, TypeScript SDK, REST API, YAML, pipelines, and MCP. The SDK supports lifecycle operations, commands, files, apps, snapshots, and configuration changes; terminal access, SSH, and custom domains are available through the UI and CLI.
What happens when a Sandbox stops?
When a Sandbox stops, compute charges stop while retained disk storage continues to be billed. A request to one of its endpoints or a command can wake it, and configured apps restart when the machine boots.
Quick Facts
- Category
- Developer tool / cloud code execution
- Platform
- Buddy cloud; isolated Linux machines
- Operating systems
- Ubuntu 22.04 LTS or Ubuntu 24.04 LTS
- Storage
- 50 GB persistent disk per Sandbox
- Resources
- From 1 CPU / 2 GB RAM up to 12 CPU / 24 GB RAM
- Management
- Web UI, CLI, TypeScript SDK, REST API, YAML, pipelines, and MCP
Alternativas ao Buddy Sandboxes
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
Manta AI
Manta AI is an autonomous web app testing tool for teams that want to map application behavior, catch regressions, and generate tests without writing scripts or maintaining selectors. It works from a URL and supports plain-English test flows, run results with screenshots, and scheduled or deployment-triggered checks.
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
Trigger.dev chat agent
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.