Persistent task chats
Each task gets its own chat rail, and the session keeps its context so it can continue after a restart instead of starting over.
Superagent
Superagent is a Mac app for coding agents with one chat per task, a real browser, streamed iPhone Simulator control, and isolated worktrees.
What Superagent is
Superagent is a Mac app for coding agents. The homepage positions it as a home for an agent, combining a chat-based workflow with a browser, persistent sessions, and device control so the agent can keep working across restarts and across different surfaces.
The product presentation centers on practical execution: one chat per task, separate worktrees for branch isolation, a real browser for logged-in work, and an iPhone Simulator that the agent can operate directly. The result is a workspace intended for coding, debugging, and page or app verification without forcing the user to babysit every step.
The homepage also indicates support for a live phone presence alongside desktop views, plus notifications when the agent finishes or needs a decision. Claude Code is shown as available today, while Codex and Antigravity are described as coming later.
Core capabilities shown on the homepage
Worktree-based branch isolation
Chats can be started on separate git worktrees so edits and commits stay on the branch for that task while the main checkout remains unchanged.
Browser automation in a live session
The agent can drive a real browser, reading, writing, clicking, and navigating pages in the same logged-in session the user already has open.
iPhone Simulator control
The product shows an iPhone Simulator streamed from its framebuffer, allowing the agent to tap, type, and swipe in a live device window.
Desktop and mobile dual view
The same page can be shown as a desktop view and a phone view at once, with changes reflected in both views during one session.
Non-intrusive notifications
When the agent finishes or needs confirmation, the interface surfaces notifications instead of interrupting the current window.
Practical ways teams can use Superagent
Isolated coding tasks
Use a dedicated chat and worktree when you want the agent to make code changes without affecting your current checkout, such as fixing bugs or moving a repo to a different package manager.
Logged-in web debugging
Use the browser-driven workflow for debugging login redirects, checking page behavior, or editing copy in a site that you are already signed into.
Mobile UI checks
Use the streamed iPhone Simulator to verify tap, type, and swipe flows in a mobile app without installing separate device tooling.
Responsive page review
Use the desktop-plus-phone layout when you need to see a launch page or responsive layout in two viewports during the same session.
Long-running agent sessions
Use notifications and restart-resistant chats when an agent is running for a while and you want to return to the session later without losing context.
Pros and Cons
Pros
- Keeps each task in its own chat with preserved context.
- Uses separate git worktrees so agent changes do not move the user’s main checkout.
- Can work in a real browser session that is already signed in.
- Includes streamed iPhone Simulator control for mobile checks.
- Shows desktop and phone views together in one session.
- Surfaces notifications when the agent needs attention instead of interrupting the window.
Cons
- The collected sources do not include pricing or plan details.
- The homepage is the main source of specifics, so setup steps, limits, and enterprise features are not documented in the provided evidence.
- Support beyond Claude Code is not available yet in the source; Codex and Antigravity are listed as coming soon.
FAQ
What kind of product is Superagent?
It is presented as a Mac app. The homepage copy also emphasizes browser-driven workflows, persistent chats, worktree-based sessions, and an iPhone simulator that can be streamed and controlled from the agent.
Which coding agents does it support?
The homepage describes Claude Code as available today, with Codex and Antigravity coming later. The source does not provide a broader compatibility list or integration matrix.
How does it organize ongoing work?
The page shows one chat per task, with each session keeping its own context and surviving restarts. It also shows sessions starting on separate git worktrees so commits land on the branch for that chat.
What can it control or automate?
The homepage says it can use a real browser, a streamed iOS Simulator, and your own logged-in browser sessions. The source does not provide setup steps or requirements beyond the Mac app framing.
Does the site list pricing?
No pricing details are available from the provided source. The /pricing URL returns a 404 page, so the site currently does not expose a usable pricing page in the collected evidence.
Quick Facts
- Product type
- Mac app for coding agents
- Primary workflow
- One chat per task with persistent context
- Automation surface
- Real browser, iPhone Simulator, desktop and phone views
- Platform
- Mac
- Source domain
- superagent.computer
- Supported agent mentioned
- Claude Code today; Codex and Antigravity coming
Superagent Alternatives
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
Ghost
Ghost is a terminal-based AI assistant for chatting, code generation, and CLI tasks. Includes free models, supports Linux, macOS, Windows, and is open source.
Vi3ecode
Vi3ecode is a maintained development environment for AI coding agents, keeping project context, terminal, Git, memory, flows and review connected.
Lucid Train
Lucid Train is a local-first AI coding harness that turns repo code into architecture diagrams for coding agents. Desktop app and Rust CLI, local models or API keys.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.