Skip to main content
You install one binary, helix. How you start it decides what kind of session you get. All three modes read your project’s AGENTS.md / CLAUDE.md, use the provider you signed in with, and share the session tools — /trace, the sub-agent views, and the rest (see Sessions).
Not sure? Start with helix. It is the full product; the other modes are subtractions from it.

​
helix — the orchestrator

You land on the Helix dashboard: the lifecycle stages, the skills that loaded, and the command roster.
The Helix boot dashboard: the gradient wordmark, the five-stage lifecycle box, the six mounted skills, setup and state panels, and the full command roster.

The dashboard helix boots into.

From here you drive the loop in plain English or with a /command — both reach the same stage:
When a stage needs parallel work, Helix dispatches sub-agents. They appear as a list under your prompt — open one and its conversation replaces the chat; typing steers that agent. That surface is described in Sessions. /help repaints the dashboard at any time; /clear starts a fresh conversation without leaving.
This is the only mode with the lifecycle commands (/spec, /build, /evaluate …). The other modes do not load the orchestrator, so those commands do not exist there — by design.

​
helix agent — a single agent as the session

No dashboard, no Helix persona, no lifecycle commands: a lean coding agent on the same harness. Sub-agents, Monitor, /goal, /trace and your project context all still work. Definitions resolve by --name (from .mutagent/agents/ and .pi/agents/), --file, --prompt, or a bare quoted prompt.

Agent mode in full

Interactive, loading a specific definition, headless with -p — plus the capability list and the sub-agent views, with real captures.

​
helix --prime — the RLM-based agent loop

Prime changes the working mode itself. Instead of the model selecting tools one call at a time, it writes and runs JavaScript against your repository, in an isolated runtime. The structure is recursive: the model decomposes a problem by writing smaller code, not smaller briefs, and when it needs help it spawns child agents from code — each of them a Prime in turn, bounded in depth. What is gone relative to helix: the orchestrator persona, the dashboard, the lifecycle skills, and the Agent tool. What is there instead: one tool, run, that executes code in a namespace that persists for the whole session — variables, functions and classes defined in one call are still defined in the next, so the model accumulates state instead of recomputing it. Every file write and shell command still passes a host gate, so “the agent ran code” still means something. Two commands belong to this mode: Combine it with agent mode to run a specific definition on the Prime loop — same agent, a code-interpreter as its only tool:
Prime is experimental.

​
The three modes side by side

Next: sessions

What the dashboard shows, the slash commands, the fleet view and the agent viewer.