Skip to main content

General

tinykit is an AI-powered development platform that lets you build, edit, and host web apps from a single deployment. Chat with AI to generate code, edit in a VS Code-like editor, and see changes live instantly.
The key difference: your builder and your app run on the same server.With other tools, you build locally then deploy separately. With tinykit, there’s no deploy step—your app is already live the moment you generate it.Plus: open source, bring your own AI key, works on mobile, multi-app support, and costs ~$10/month total.
tinykit is in early beta. Great for:
  • Personal utilities and side projects
  • Internal tools and admin dashboards
  • MVPs and prototypes
Keep in mind:
  • No guarantees against data loss—back up anything important
  • Auth requires manual wiring (no built-in user auth for generated apps)
  • Best suited for internal use or tools where you control access
Not recommended (yet) for public-facing apps with sensitive user data.
Yes! MIT licensed. Fork it, customize it, self-host it, sell services with it. View on GitHub

Deployment & Hosting

No. Railway is the easiest option (one-click deploy), but you can host anywhere that runs Node.js:
  • Fly.io
  • DigitalOcean
  • AWS
  • Any VPS
  • Docker
See Architecture for deployment guides.
Yes! tinykit supports domain-based routing. Point multiple domains to your server and each serves a different app:
recipes.yourserver.com → Recipe app
blog.yourserver.com    → Blog app
crm.yourserver.com     → CRM app
Run hundreds of apps from a single tinykit instance.
Yes! Your apps compile to standalone HTML. Export and deploy anywhere:
# Download your app from the builder
# Deploy to Vercel, Netlify, GitHub Pages, etc.
Yes. Configure your custom domain in Railway (or your hosting platform) and point your DNS there. tinykit automatically serves the right app based on the domain.
Railway and most modern platforms provide automatic HTTPS. No configuration needed.

AI & API Keys

  • OpenAI - GPT-4o, GPT-4 Turbo, GPT-3.5
  • Anthropic - Claude Sonnet 4, Claude 3.5 Sonnet
  • Google Gemini - Gemini Pro, Gemini Flash
  • z.ai - GLM models
This keeps costs transparent and in your control. You pay your AI provider directly for what you use—no markup, no hidden fees, no subscription.
Typical casual usage: $3-5/month with GPT-4 or Claude.Heavy usage (building apps daily): $10-20/month.You can monitor and set limits in your provider’s dashboard.

Features & Usage

Yes! The editor and AI work great on tablets. Perfect for quick edits on the go.
The built-in Monaco editor is powerful, but you can also:
  • Clone the workspace to your local machine
  • Edit with VS Code, Cursor, or any editor
  • Push changes back
The AI agent works best through the built-in interface though.
tinykit generates Svelte apps by default, compiled to standalone HTML:
  • Svelte 5 with runes
  • Tailwind CSS
  • PocketBase for data
The AI can generate other code too—you just need to configure how to run it.
Yes! Snapshots are built in. Every AI change creates a snapshot you can restore to. It’s like time travel for your app.
tinykit includes PocketBase—an embedded database. Create collections, store data, set up real-time subscriptions. The AI can wire up your data automatically.

Troubleshooting

Check:
  1. Your API key is correct in .env
  2. You have credits/quota with your AI provider
  3. The model name is correct (e.g., gpt-4o, not gpt4)
Check your provider’s dashboard for usage and errors.
Try:
  1. Hard refresh the preview (Cmd+Shift+R)
  2. Check the browser console for errors
  3. Ensure files are saving (check for save indicator)
Check:
  1. File size is under 5MB
  2. File extension isn’t blocked (no .exe, .sh, etc.)
  3. Path doesn’t contain ../
Check browser console for specific error messages.
Check:
  1. Your domain’s DNS is pointing to your tinykit server
  2. The domain is associated with a project in the dashboard
  3. SSL/HTTPS is properly configured
Visit /tinykit/dashboard to see all projects and their domains.

Still Have Questions?