Skip to main content
helix boots into the orchestrator session. The other modes share the same harness and most of the same commands — Three ways to run Helix notes what each drops.

The dashboard

helix opens on the dashboard: the lifecycle stages, the skills that loaded, and the command roster. It is a header, not a menu — you talk to Helix in the prompt beneath it. /help repaints the full dashboard into the transcript at any time, without a model turn.

Plain English or a slash command

Describe what you want, or type the command — both route to the same stage:
/commands complete as you type. Underneath, each one wires to the *command the dashboard roster shows and Helix uses in its own replies — /diagnose and *diagnose are the same thing. The full roster is on Commands.

Session commands

The crew

Sub-agents Helix dispatches appear as a list under your prompt; open one and its conversation replaces the chat, and typing steers it. The full surface — the fleet list, the viewer, the keys — is on Sub-agents.

The working line

While the model works, the line under the transcript shows elapsed time, tokens, and what it is doing (thinking escalates to overthinking after a minute, and settles to thought for Ns when reasoning ends). An open agent viewer shows the same line, on that agent’s clock.

Next: commands

The lifecycle roster, the session commands, and what to type on which surface.