New: Multi-agent support — Claude, Codex, and Gemini in one workspace →

A To-Do List For Agents.

Hand off tasks to Claude, Codex, or Gemini—and watch them ship.

Task-first. Terminal-native.

A workflow designed for how developers actually work.

1

Create task

Add a task with ⌘N. It automatically gets its own terminal session.

2

Work in terminal

Run commands, install deps, edit files. Your shell state persists with the task.

3

Launch agent

Send to Claude, Codex, or Gemini. The agent runs in your task's terminal with full context.

4

Ship

LED turns green on commit. Mark complete. Your terminal history is saved forever.

Built for power users.

Every feature designed for developers who demand more from their tools.

Select Agent
C
Claude Code
Opus · Sonnet · Haiku
O
OpenAI Codex
Full Auto mode
G
Gemini CLI
Interactive mode

Multi-Agent AI

Choose your AI. Claude Code, Codex, or Gemini CLI—each with multiple execution modes. Switch agents per task.

Local Whisper transcription

Voice Input

Speak your tasks. Built-in Whisper transcription runs locally on your machine. Privacy-first, hands-free task capture.

Idle No activity
Working Agent processing
Clarification Agent asking
Permission Needs approval
Complete Ready for review

LED Status System

Know exactly what's happening at a glance. 5-state LED indicators show agent status in real-time—no more checking in.

TASK-42: Add auth
~/project/src
$ npm install passport
added 12 packages
$ git status
modified: package.json
Session persisted Saved

Persistent Terminals

Every task gets its own shell session. Close the app, reopen—your terminal state, history, and working directory are exactly where you left them.

Keyboard-first
New task ⌘ N
Navigate
j k
Focus terminal ⌘ ⇧ ↵
Jump to prompt
⌘ ↑ ⌘ ↓

Keyboard Shortcuts

Navigate entirely from the keyboard. Jump between tasks, focus the terminal, search—all without touching your mouse.

screenshot-error.png
Passed to AI agent
design-mockup.png
Reference image

Drop images or paste from clipboard

Image Context

Attach screenshots to tasks. Paste from clipboard or drag and drop. Images are automatically passed to AI agents for visual context.

Your keys. Your models.

Connect directly to Claude, Codex, or Gemini with your own API keys. No middleman, no markup, no extra subscriptions.

Direct API Access

Connect straight to Anthropic, OpenAI, or Google.

No Platform Fees

Pay only what you'd pay the providers directly.

Switch Anytime

Swap agents per task. No lock-in.

Use Your Credits

Existing API credits work immediately.

How it works

AI Provider Your choice
API Billing Direct to provider
Model Access Full catalog
Rate Limits Your account's limits
Data Retention Provider's policy

Free

Everything you need, bring your own keys

$0
  • Unlimited tasks & projects
  • Persistent terminal sessions
  • Multi-agent support (Claude, Codex, Gemini)
  • LED status system
  • Voice input (local Whisper)
  • Image context support
Download free

Tasks and terminals. Finally unified.

Download Rithm free for macOS. Windows and Linux coming soon.

macOS

Available now

Windows

Coming soon

Linux

Coming soon

Download for macOS

No sign up required