
Sign in to rate this tool.
Overview

Orb is a platform that brings together multiple AI providers, machines, and agent uses under a single interface. It works on macOS, iOS, and Linux servers, letting you connect subscriptions, API keys, and local hardware to run agents such as Claude Code, Codex, Antigravity, OpenCode, and Grok. The service originated as sandboxed.sh and now serves as the backend for Orb clients.
The tool is aimed at developers and teams that need to orchestrate AI agents across different environments. You can launch an agent locally on your Mac using Orb's Tauri runner, on a private Linux server or fleet node, or through cloud coordinators like Hermes, Grok Bot, or Cursor Cloud. Each execution mode separates where you read and steer the conversation from where the agent actually runs, allowing you to keep project files, logs, and mission transcripts synchronized to a central server. Projects are organized with per‑project colors, a Skills folder containing markdown definitions, and a bi‑directionally synced @context folder that holds shared files.
Using Orb involves a few steps: connect your AI provider accounts (Anthropic, OpenAI, xAI, Kimi, Google Antigravity) via the Providers panel, configure a machine or cloud node, and start a new agent from the composer. Inside a conversation you can expand collapsible activity folds, view nested sub‑agents, and switch between /plan mode (question and approval cards) and /goal mode (autonomous multi‑turn execution). A side‑agent panel (⌘⇧J) lets you ask side questions without interrupting the main turn. Missions can be scheduled to wake up later, automatically recover from interruptions, and be transferred between machines with SHA‑256‑verified file manifests and Git bundles. Output supports markdown, syntax‑highlighted code, tables, KaTeX math, interactive quizzes, inline images, and clickable workspace file links.
On iOS, Orb mirrors the macOS experience: you can browse projects, inspect mission transcripts, send follow‑up instructions, edit shared markdown context, and manage backend machines and providers. The iOS client also shows unread activity and lets you jump between active repositories and controllers.
Key features
- Connects multiple AI providers and local hardware in one workspace
- Three execution modes: local Mac runner, private Linux node, cloud coordinator
- Project organization with colors, Skills markdown, and synchronized @context folder
- Interactive conversation UI with collapsible folds, /plan and /goal modes, and side‑agent panel
- Mission scheduling, automatic recovery, and cross‑machine transfer with SHA‑256 verification
Details
- Pricing
- Free
- Category
- Code
Suggest an edit
Sign in to suggest an edit.
Pros and cons
Pros
- Supports a wide range of agents and providers in a single interface
- Allows smooth switching between local, private, and cloud execution environments
- Provides detailed UI controls for planning, goal execution, and side queries
- Synchronizes project files and mission transcripts across macOS, iOS, and server backends
Cons
- No pricing information is published on the site
- Requires manual setup of servers or Docker for backend deployment
- Documentation references many features but does not include step‑by‑step tutorials for all execution modes
- Limited to macOS, iOS, and Linux; no native Windows client is mentioned

