LiveSplit-style progress tracker for MoSim robot mods. Sign in with Google → your data syncs across every device (browser, desktop app, phone).
Tracks robots through Planned → In Unity → Semi-Functional → Released, with
10 steps / 53 sub-steps derived from the official MoSim docs
(modding intro + 9496 Lynk walkthrough).
Edit steps.json at the repo root to change your workflow.
steps.json Shared split template (web + desktop both read this)
web/ React + Vite frontend (runs in browser and inside the Tauri app)
src-tauri/ Rust Tauri shell — builds the native desktop app
server/ Express REST API + SQLite + Google OAuth
All data is stored in a SQLite database (server/data.db) on your own server.
No Firebase, no third-party database service.
| Guide | Contents |
|---|---|
| Server Setup | Deploy the server to a DigitalOcean droplet with nginx + HTTPS |
| App Installation | Install the desktop app on Windows, macOS, or Linux |
| AI Model Training | Fine-tune a local Ollama model on your script library |
| Development | Run everything locally for development |
- Robots — spreadsheet-style tracker: Status/Mod Type pills, modpack, repo link, progress bar, notes. Filter chips across the top. "In Unity" rows show the current step. Click any row → splits view.
- Robot detail — 10 collapsible steps with 53 checkable sub-steps, a note box per step, "What's left" panel, overall notes, and the AI script generator.
- Planned — robots you haven't started yet. "Start modding →" moves one to In Unity.
- Modpacks — the Addressables packs your robots ship in.
- Repos — your git repos. The desktop app can Scan a repo to find every robot folder.
- Scripts — your personal
.cslibrary. Drag files in; the AI generator feeds all of them as examples. Export as JSONL for fine-tuning (see TRAINING.md). - Compact — LiveSplit-style overlay column.
/#/compactin any browser, or what the desktop app shows by default.
Themes: 🌙/☀️/☁️ in the top bar cycles Dark → Light → Cloud (soft blue).
On any robot's page, the AI panel lets you describe the robot and generate a complete C# script using the real MoSim APIs.
Two providers:
| Provider | Cost | Internet |
|---|---|---|
| Claude API (Anthropic) | ~$0.01–0.10/script | Yes |
| Local Ollama model (your own fine-tune) | Free | No |
Your API key and Ollama settings are stored only in that device's localStorage —
never written to the server.
See TRAINING.md for how to fine-tune your own model on your script library.
A native app built with Tauri (Rust + local WebView). The UI renders from a bundled local build — no loading screens, no dependency on the server being up to see the interface.
Download the latest installer from the Releases page and see docs/app-install.md for platform-specific instructions.
- Starts in compact mode (360 × 640, always on top) — the LiveSplit overlay
- 📌 toggles always-on-top. ⛶ expands to full UI. ✕ closes
- From the Repos page, Scan reads your local git repo folders for robot prefabs
During first setup the installer asks for your server URL. You can change it later by
editing mosim.conf next to the app executable.
See docs/server-setup.md for a full step-by-step guide covering:
- Creating Google OAuth credentials
- Creating and configuring a DigitalOcean droplet
- Automated setup via
server/manage.sh - nginx reverse proxy + subdomain + free HTTPS via certbot
Quick deploy after initial setup:
sudo bash server/manage.sh deploy # git pull + rebuild + restart
sudo bash server/manage.sh logs # tail live logsEvery robot, modpack, and repo has a Private toggle. The server only returns your own data — all endpoints require sign-in and filter by your Google account.