Multi-agent team
Run CEO, Coder, QA, DevOps, PM, and Researcher roles as first-class collaborators.
Archive — Tron OS V1. This is the original Tron OS interface, kept for reference.
The current design is Tron OS V2.
A note on AI: Sleepless Games does not — and likely never will — use AI in the final products of any of our games. Side projects like Tron are just that… side projects — but as a solo developer, I have to pay the bills somehow.
Sleepless Games AI command center
Tron OS V1
Your AI Agent Hub — a local command center that drives the official AI CLIs as a team.
Run humans and agents from one mission board: projects, chats, memory, news, actions, and companion control from Android. Tron OS is the hub, not a harness.
CEOplanning
Codershipping
QAverifying
DevOpsstanding by
OpenClawCLI linked
A command deck for every run.
Desktop for deep work. Your phone for remote decisions — Android now, iOS on the way. Both keep the team moving through the same mission control surface.
https://tron-os.local
Tron OS on your desktop
9:41Tron Remote
Mission Control
Team Room live
The Dark Is Sleepless funding push
3 active agents, 2 runs waiting on review, one supporter post queued.
Action Layer
Approve, pause, comment, or hand a task back to the team from your phone.
@Coder tighten the companion copy.
Coder: draft ready for QA.
@QA smoke test the supporter flow.
Get the appGet it on Google PlaySideload APK
9:41Tron Remote
Coming soon
iOS
Tron OS Remote for iPhone & iPad is on the way.
iOS — coming soon
Everything the AI team needs in one hub.
Tron OS connects the official CLIs into a practical workspace for shipping, reviewing, and coordinating work.
Project pipeline + Kanban
Track goals, issues, blocked work, review states, and handoffs from a single board.
Memory galaxy
Keep durable context visible so agents can work with history instead of starting cold.
News Room
Watch product, studio, and market signals without leaving the operating surface.
Team Room chat
@mention an agent, route a decision, or pull a specialist into the thread.
Action Layer
Approve launches, request changes, pause work, and keep execution auditable.
Android companion app
Carry mission control in your pocket for approvals, chat, alerts, and remote triage.
Official CLI hub
Drive Claude, Codex, Hermes, and OpenClaw from one command center. It is a hub, not a harness.
See it in action
A look inside Tron OS — the command center, your AI company, and the tools they run. Swipe or use the arrows.
Everything packed in
One local command center for your whole AI operation. Scroll the full feature set.
🔗
GitLab & GitHub
Connect either code host — or both. Your AI staff watch your issues and merge / pull requests and act with your approval.
👔
Staff Triage
Your AI company watches your boards, triages bugs and feature requests, drafts fixes and posts, and proposes — nothing ships until you approve it.
🔀
Merge & pull requests
Review, approve and merge MRs and PRs from the dashboard. An approved fix becomes a Coder-built request, ready for your merge.
🔔
Notifications everywhere
Get pinged in the OS, in your browser, and on the Android app — wherever you are, you never miss an approval or an event.
🔐
Team login & roles
Multi-user accounts with roles and per-feature permissions, layered on top of secure device pairing.
💬
Team & agent chats
A shared team room plus 1:1 chats with every agent — each a different real model — all synced across your devices.
🐛
Built-in feedback
One-click bug reports and feature requests from the dashboard and the app route straight to your issue tracker.
⬆️
Updates — manual & auto
One-line OS updates, plus in-dashboard updates for the agent CLIs (Claude, Codex, OpenClaw, Hermes) that restart themselves.
📎
Artifacts
A shared file space for you, every agent, and your team — drop a file, a zip, or a plugin and hand it off.
⚙️
Deep settings
Toggle agents and features, manage outbound connections, paired devices, the voice assistant, and far more.
📰
News Room
Give the team a topic; on your schedule it scrapes the web — Google, Reddit, HN, X, YouTube — and a free model writes the briefing.
🧠
Memory galaxy
Every note and memory, searchable, in a living star map your agents read and write.
🛰️
Hivemind
A live group chat where all your agents riff off each other, read your vault, and can start real work.
🗂️
Pipeline + Kanban
Capture an idea → it gets classified, routed, planned and built — you approve once and watch it happen.
📱
Android companion
Mission Control in your pocket. Pair once and the app inherits everything the server can do.
🎨
Make it yours
A theme engine for colours, pane layout and navigation — plus an embedded ComfyUI and Jarvis voice.
📚
Asset Library
One searchable library across every studio — images, thumbnails, video, audio, SEO and games — each with its provenance, ready to reuse and export.
📥
Review & Approval
One inbox to approve, retry and share everything the fleet produces — renders, tracks, SEO drafts, commissions — from your desk or your phone.
🚩
Launch Campaigns
Turn one topic into a coordinated launch — a checklist that walks your studios in order: News → SEO → video → thumbnail → music → game → publish.
🎯
Agent Missions
State a goal; Tron breaks it into tasks, routes each to the agent best suited to it, and tracks per-agent progress to done.
📊
Cost & ROI
Per-artifact cost, time and model — broken down by studio, campaign and agent, so you see what every run really costs.
🎬
Creative studios
Video, music, podcast, thumbnails, SEO and playable games — each a studio the fleet runs from a single brief to a finished file.
🎮
Game development
Agentic Unreal Engine world-building — describe a place and the fleet builds it in-editor, then drives your project from the dashboard.
🚀
Game publishing
A Steamworks hub for your app, builds and depots — and it generates the SteamPipe CI that ships your releases.
🧩
Model Hub
Pull models from Ollama or HuggingFace, switch the active model, and give any agent its own — all running locally on your GPU.
🏠
Smart home & security
Drive Home Assistant and route a voice speaker to Tron's brain — plus UniFi devices, clients and live camera snapshots.
📅
Calendar
One calendar across Google, Microsoft, CalDAV and ICS, synced both ways, shared with the agents that schedule your work.
🔌
Connector Hub
One place to wire the OS into the services you already use, with credentials that stay on your machine.
🖥️
Server Manager · soon
Tiles for every machine you run — Hetzner, DigitalOcean or a plain SSH box — with agents installing software, opening ports and hardening SSH.
📡
Autopilot
A unified health map of the whole system — what's running, what needs attention, and what the fleet is doing right now, at a glance.
🛡️
Output Watchdog
Watches the fleet's output for problems and flags them before they ship — a quality tripwire on everything the agents produce.
🎞️
Flight Recorder
A black-box recorder of everything the system does — replay any run, trace what happened, and diagnose issues after the fact.
↕ scroll for more — and new features land every week.
Install in one line
Tron OS is local-first and self-hosted. Paste this into a terminal on your own machine — no sudo, and by default nothing leaves your computer.
curl -fsSL https://tron.sleeplessgames.co/install.sh | bashSame one-liner on Linux and macOS — the installer auto-detects your platform. (macOS has no background service, so after it finishes start the dashboard yourself with cd ~/tron-os/source && npm run start.)
What to expect: Linux or macOS (x64/arm64). It runs unattended for a few minutes — fetching the dashboard, building it, and (on Linux) registering a boot-safe background service. When it finishes it prints a link; open http://localhost:3737 and you’re in. Re-running it later just updates in place.
✓ What it sets up
- The Tron OS dashboard in
~/tron-os, built and served on localhost:3737 (and your LAN). - A private, local copy of Node.js — only if your system doesn’t already have a recent one.
- A boot-safe background service plus a
tron-oscommand (start / stop / update / configure). - Empty, locked-down API-key placeholders under
~/.agentic-os— every paid feature stays dormant until you add a key. - Optional, on request: the local agents, a headless Obsidian vault, and a daily update check.
✗ What it won’t do
- No sudo, no root. Everything lives in your home folder — system packages and system Node are left alone.
- No keys bundled, nothing phones home on its own. AI providers stay dormant until you add your own key — but once you connect a cloud agent, the prompts you send it go to that provider (see the note below).
- Not exposed to the internet. It binds to your machine and LAN only — no public ports opened, no domain set up for you.
- Won’t overwrite your existing config or keys, and won’t auto-update unless you opt in.
- No native Windows — but it runs on Windows through WSL (Windows Subsystem for Linux): install WSL, open your Linux shell, and run the same one-liner there.
Privacy note: Tron OS is local by default. If you choose to connect an internet-based AI provider (Claude, OpenAI / Codex, Gemini, and the like), the prompts and content you send to that agent are transmitted to that provider to be processed — exactly as with any app built on their API. For fully on-device operation, run local models (e.g. Ollama) instead.
Support Tron. Support The Dark Is Sleepless.
Some features are supporter-only. Tron runs on the SAME Patreon as Sleepless Games’ survival game "The Dark Is Sleepless" — we added the Tron AI Agent Hub to the pipeline to help BOOST the game’s pre-announcement funding. Supporting Tron directly supports The Dark Is Sleepless.
Back the studio, unlock the supporter path, and help fund the tooling that keeps the game moving.