AI-native client-only browser platform for designing, programming, wiring, simulating, and deploying embedded systems. No backend — the browser talks to Turso (libSQL) and AI providers directly.
| Layer | Stack |
|---|---|
| App | React 19, Vite 8, TypeScript 6, TailwindCSS 4, Zustand, TanStack Query, Monaco Editor, Framer Motion |
| Data | Turso (libSQL/SQLite) over HTTP via @libsql/client + Drizzle ORM, called directly from the browser |
| Simulation | Custom C++ parser → IR → VM, netlist solver, device SDK, running in a Web Worker |
| AI | Bring-your-own-key, OpenAI-compatible providers (OpenAI, Anthropic, Gemini, OpenRouter, custom) called directly from the browser |
⚠️ Security note: Because there is no server, the Turso auth token is bundled into the client and exposed to users. Ownership is enforced at the application layer only (queries filter byuserId). Use a token with minimal permissions. Seespec/RFC0001for the full tradeoff analysis.
| Landing | LED Blink | Servo Sweep |
 |
 |
 |
rakit-io/
├── src/
│ ├── components/ # UI components
│ ├── pages/ # Route pages (Landing, Auth, Dashboard, Editor)
│ ├── lib/
│ │ ├── db/ # Drizzle schema + libSQL (Turso) browser client
│ │ ├── svg/ # SVG board & component renderers
│ │ ├── simulator/ # C++ parser, IR, VM, netlist, device SDK
│ │ ├── stores/ # Zustand stores (auth, project, simulation, ui)
│ │ ├── hooks/ # TanStack Query hooks
│ │ ├── api.ts # Data + AI access layer (Turso / providers)
│ │ └── types.ts
│ └── workers/ # Simulation Web Worker
├── scripts/
│ ├── migrate.ts # Idempotent schema creation
│ └── seed.ts # Demo user + example projects
└── spec/ # Architecture RFCs
bun installcurl -sSfL https://get.tur.so/install.sh | bash
turso auth login
turso db create rakit-io-db
turso db show rakit-io-db --url # connection URL
turso db tokens create rakit-io-db # auth tokenCopy .env.example to .env and fill in your Turso credentials (and an
optional default AI key):
VITE_TURSO_URL=libsql://<your-db>.turso.io
VITE_TURSO_AUTH_TOKEN=<your-token>
VITE_AI_BASE_URL=https://api.openai.com/v1
VITE_AI_API_KEY=
VITE_AI_MODEL=gpt-4o-miniexport TURSO_URL="libsql://<your-db>.turso.io"
export TURSO_AUTH_TOKEN="<your-token>"
bun run migrate # create users, sessions, projects, ai_providers
bun run seed # demo user + 8 example projectsbun dev # Vite dev server on http://localhost:5173| Field | Value |
|---|---|
demo@rakit.io |
|
| Password | demo1234 |
The demo account comes pre-seeded with example projects: LED Blink, DHT22 Weather Station, Servo Sweep, Button LED, Potentiometer, Servo Knob, BME280 Monitor, OLED Hello World.
bun dev # Start Vite dev server
bun build # Production build (tsc + vite)
bun typecheck # Typecheck
bun lint # Lint with oxlint
bun migrate # Create/ensure DB schema
bun seed # Seed demo data
bun deploy # Build and deploy to Cloudflare Pages
bun pages:dev # Build and run locally via Wrangler PagesThe build output in dist/ is a fully static SPA deployed to Cloudflare Pages
via Wrangler. wrangler.toml points Pages at dist/, and
public/_redirects ensures client-side routes fall back to index.html.
-
Set build-time environment variables in the Cloudflare Pages dashboard (Settings → Environment variables) for both Production and Preview, or pass them on deploy. These are baked into the static bundle:
VITE_TURSO_URL=libsql://<your-db>.turso.io VITE_TURSO_AUTH_TOKEN=<your-token> VITE_AI_BASE_URL=https://api.openai.com/v1 VITE_AI_API_KEY= VITE_AI_MODEL=gpt-4o-mini
-
Authenticate once:
npx wrangler login
-
Deploy:
bun deploy # or, with inline env vars for a one-off: npx wrangler pages deploy dist --var VITE_TURSO_URL:... --var VITE_TURSO_AUTH_TOKEN:...
- Arduino Uno (Nano, Mega), ESP32 DevKit V1, ESP8266 NodeMCU, Raspberry Pi Pico
LED, Button, Resistor, Potentiometer, DHT22, BME280, SSD1306 OLED, Servo, RGB LED, Ultrasonic, LCD 16x2, Buzzer, Relay.
MIT