phasor discovers every running AI coding agent — the ones it launches in tmux and the ones you started anywhere else — and gives you a single live view to watch, open, and steer them. In the terminal, or your browser.
Finds every running agent process automatically and shows each as its own live node — title, status, todo progress, activity, touched folders.
Agents launched from phasor live in its own tmux session and open with one key. Agents started elsewhere are discovered too — monitored, never disturbed.
Map directory prefixes to a name and colour. Matching agents are grouped, tinted, and labelled — editable from both the terminal and the browser.
Queue a prompt to fire the moment an agent finishes its turn — repeated until the work is genuinely complete. Chain steps without babysitting.
Drive any managed agent's real terminal live in the browser over a WebSocket bridged to its session. Type, resize, detach — bound to localhost.
A per-agent load meter shows how hard each is working; finishing a task is marked in the terminal and bursts in the graph.
◍ phasor 1 in tmux · 6 external ╭ ◆ Forge ───────────────────────────╮ │┏━┓ ● Analyze the build pipeline now│ │┗━┓ ▰▰▰▰▰▱ 5/8 ⚡42%│ │┗━┛ Build and test:… │ ╰───┬────────────────────────────────╯ ├──▶ api ├──▶ parser ╰──▶ runtime
# needs tmux + an agent CLI on PATH, and Rust to build cargo build --release # terminal dashboard ./target/release/phasor # browser dashboard phasor serve # http://127.0.0.1:7878 # spawn an agent in any project cd /path/to/project && phasor exec claude
| phasor | the dashboard (TUI) |
| serve [port] | web dashboard, default 7878 |
| exec CMD… | run CMD in a new window, then exit |
| start CMD… | run CMD and open straight into it |
| save [file] | snapshot all sessions (managed + external) |
| restore [file] | recreate & resume saved sessions |
| doctor [cwd] | diagnose transcript resolution |
| render [WxH] | print one frame as plain text |