Frequently asked questions

What is LongHorn?

LongHorn is an open-source, modular TypeScript backend framework built on Hono and Web Standards. One application runs on Node, Bun, Deno or Cloudflare Workers, and you import only the modules you use, such as database, auth, queues, schedules and outbox delivery.

How is LongHorn different from Hono?

LongHorn builds on Hono rather than competing with it: app.http is a real Hono app, so Hono routers and middleware work directly. Hono handles HTTP, and LongHorn adds application lifetime, dependency injection, native hosts and modules for storage, auth, jobs and delivery. Full comparison.

Can I run LongHorn on Cloudflare Workers?

Yes. createWorker from @longhornjs/workers exports your app as a Worker, with D1 and Durable Object SQLite for storage. Some things differ: there is no Worker scrypt provider yet, so password auth doesn't run there, and the complete console application is verified on Node only.

Can different services run on different runtimes?

Not as a framework feature yet: per-service runtime placement is on the roadmap. Today each application picks one host, and separate applications can call each other over HTTP.

Is LongHorn production-ready?

No. LongHorn is an early alpha: it has not been deployed to production and its APIs can still change. It is on npm, but that is a first release, not a stability promise. Its failure handling is tested heavily, with 749 source tests and 66 verification gates, but that is not the same as production use.

Is LongHorn the Laravel for TypeScript?

Partly. Like Laravel it offers auth, queues, schedules, caching and validation as first-party modules, but it has no ORM, generated CRUD or admin panel, and nothing starts unless your application imports and wires it.

How do I install LongHorn?

Run npm create longhorn@latest my-api -- --runtime node, then pnpm install and pnpm dev inside the new folder. You need Node 20.19 or newer and pnpm 12.6. Getting started.

What does LongHorn test that other frameworks might not?

LongHorn tests failure by causing it: processes killed with SIGKILL during and after commits, lost HTTP replies, concurrent writers and failed schema upgrades, against real PostgreSQL, Redis, D1 and Durable Objects. Delivery is at-least-once with deduplication, not exactly-once across systems.

Does LongHorn include AI or LLM features?

No. LongHorn has no built-in LLM features, by design. It is a backend framework, and you can call any AI service from your own code.

What license does LongHorn use?

LongHorn is MIT licensed. The source is at github.com/MirHamzaHasan/longhorn.