Getting started with BATCH

BATCH is a no-build, server-rendered hypermedia substrate: Bun runs the TypeScript directly (even client .ts is transpiled on request), there's no bundler, and htmx handles reads/nav. This page is the fastest path from a clone to a running app; the reasoning behind every decision here is ARCHITECTURE.md, and the build rules are CONVENTIONS.md.

Install and run

bun install            # or: npm install (bun came in via npm here)
bun run dev            # http://localhost:3000 — hot reload, no build step
bun run check          # tsc --noEmit — must stay green
bun test               # unit + integration

That's the whole loop: edit a file, refresh the browser. There is nothing to compile or bundle between your source and the server.

What you get out of the box

  • The composition engine — server-rendered HTML from .html templates + a binding vocabulary (data-field, data-bind-<attr>, each, component tags) — see CONVENTIONS §2–4.
  • A generic SSE push hub (batch/http/stream.ts) — the transport GRAIN's render ops ride on.
  • Sitemap, robots.txt, and /llms.txt generated from one route source — SEO/AEO for free.
  • A static exporter (batch/export) that freezes the running server's output — never a second renderer (ARCHITECTURE §18).
  • A framework-generic perf + SEO/AEO audit (bun run audit).

Next steps

  • Read ARCHITECTURE.md for the substrate's full reasoning (start here if you want to understand why, not just how).
  • Read CONVENTIONS.md for the component/layering/testing rules before you add code.
  • If you also want the AI-interaction layer (a UI a human and an AI can operate through one vocabulary), see GRAIN's getting-started — it builds on BATCH but imports nothing from it.
  • Browse every component live at /catalog.