Mutsumi

v0.4.0b1 Beta

You don’t need focus. You need to never lose a thread.

A dozen contexts a day — coding, reviewing, messaging, running agents, scanning feeds. That’s not a flaw. That’s your operating mode. The problem is: the moment you switch away, the last thread starts fading. Mutsumi keeps all your threads visible — summoned in a keystroke, gone in another.

┌─────────────────────────────────────────────────────┐
│  [* Main] [Personal] [proj-a] [proj-b]   mutsumi ~  │
├─────────────────────────────────────────────────────┤
│  v HIGH ─────────────────────────────────────────   │
│  [ ] Refactor Auth module             dev,backend ***│
│  [x] Fix cache penetration bug        bugfix      ***│
│  v NORMAL ───────────────────────────────────────   │
│  [ ] Write weekly report              life        ** │
│  [ ] Review PR #42                    dev         ** │
├─────────────────────────────────────────────────────┤
│  6 tasks · 2 done · 4 pending                       │
└─────────────────────────────────────────────────────┘

Why Mutsumi?

Peripheral Vision

Not center stage. Not hidden. She lives at the edge of your screen — one glance tells you every active thread and what your agents have pushed forward.

Agent-Driven

Claude Code, Codex CLI, Gemini CLI, Aider, or a shell script — your agents write tasks for you. You just glance and confirm.

Summon & Dismiss

One hotkey to summon. One hotkey to dismiss. Quake-mode terminal, tmux popup, or a tiling split — she appears when you need her, vanishes when you don’t.

Hackable

TOML config, custom themes, custom keybindings, Textual CSS overrides. Mod everything. Mutsumi loves being customized.

How It Works

graph LR
    A["AI Agent<br/>(Controller)"] -->|"write"| T["mutsumi.json<br/>(Model)"]
    T -->|"watch"| V["You (TUI)<br/>(View)"]
    V -->|"interact"| T

Your agents write JSON. Mutsumi watches and renders. You glance, confirm, and move on to the next thread.

Quick Install

uv tool install git+https://github.com/ywh555hhh/Mutsumi.git
mutsumi