Runix turns intent into autonomous execution, live driver action, and reusable artifacts.

Move through four architecture views while the runtime mesh reacts to real output in the console. This page is not a static explainer. It is the operating surface for a system that routes intent, drives software, and records the run.

github capability matrix

Live telemetry Phase 01 · Engine map

Health Booting

Drivers --

Actions --

Steps --

Runix starts with orchestration: one request enters the engine, then the engine selects the next driver or workflow step.

Runix live console

Ready

Uptime --

Version --

Drivers --

Actions --

Voice Console

Web Speech API — click the mic, speak. Toggle wake word to say "hey runix" anytime.

You

—

Runix

—

Audio output here uses browser speech synthesis. Native Runix voice and wake training use ElevenLabs outside this hosted page.

AI Chat — real Runix hosted chat

Authenticated only. Simple questions use hosted Ask Mode; web-dependent questions are promoted to the hosted AI agent so it can browse before replying.

Ask anything. This card runs the hosted Runix engine directly, and it can use the live web for external questions when needed.

🎛 Runix Command Center — full CLI surface

Every runix command, with form inputs for its arguments. Pick a category, fill in the fields, click ▶ Run. Raw stdout streams into the terminal below — always visible.

🔒 Sign in to unlock arbitrary runix commands (ai, run, driver, schedule, artifacts, checkpoints, resume, voice).

One-Click Live Workflows

Shortcut library — each card hands a live prompt to runix ai agent. Output streams in the terminal below while the hosted engine works the task.

Idle — click any ▶ Run button above

Interval idle

LIVE

Click ▶ Start live preview to mirror your desktop here. Web automation runs in headed mode — Chromium windows will appear as they open.

Safe live tasks route only to read-only commands the deployed server allows right now. Sign in to run arbitrary engine commands.

Public sandbox: only deterministic example-driver steps (echo, add, wait). Sign in to unlock arbitrary Gherkin against the live engine — AI, web, system, vision, office, and more.

Every driver is an independent WebSocket server with its own driver.json manifest. The engine discovers them at runtime — adding a driver means dropping a folder and restarting.

Every action is addressable as a Gherkin step. The engine matches Cucumber expressions ({string}, {int}) and a universal I call driver.action(args) form.

The full capability matrix. This is what ships — not just what the public sandbox exposes.

Every feature, capability, driver action, CLI flag, MCP tool, and integration compiled from the Runix codebase — loading…. Everything Runix does, in one list.

Recent runs are saved locally in the browser. Click any entry to replay it in the live task tab.