"Because going faster shouldn’t mean writing more nonsense"

---

OxarionJs is a TypeScript-first backend framework built on Bun.

It keeps the flow direct, the API clean, and the abstractions where they actually help.

Why OxarionJs

• Built on Bun for fast runtime and a tight development loop
• Type-safe route params
• Route groups and middleware chains
• Route injector for modular structure
• File-based dynamic routing
• Request and response helpers
• Native WebSocket route support
• Cached SSR templates and optional Ox dynamic HTML runtime
• Minimal flow without high-level clutter

Install

bun add oxarionjs

Quick start

import Oxarion, { Middleware } from "oxarionjs";

await Oxarion.start({
  host: "127.0.0.1",
  port: 3000,
  httpHandler: (router) => {
    router.addHandler("GET", "/", (_req, res) => {
      res.json({ message: "Welcome to OxarionJs" });
    });
  },
  safeMwRegister: (router) => {
    router.multiMiddleware("/", [Middleware.cors(), Middleware.logger()], true);
  },
});

Run it with

bun run src/index.ts

Dynamic routing

import Oxarion from "oxarionjs";

await Oxarion.start({
  dynamicRouting: {
    dir: "dyn",
    handlerFile: "api",
    extensions: ["ts", "js"],
    onConflict: "keepManual",
  },
  httpHandler: () => {},
});

dyn/test/api.ts maps to /test.

Route modules can export HTTP method functions or a static class.

import type { OxarionRequest, OxarionResponse } from "oxarionjs";

export default class TestApi {
  static async GET(req: OxarionRequest, res: OxarionResponse) {
    return res.json({ path: req.url() });
  }
}

Benchmark

OxarionJs is built to stay lightweight on the hot path, and local benchmarking reflects that well.

Here is an example result from the Bun benchmark suite used during development:

Framework	Average Req/Sec
Oxarion	232632
Previous baseline	179348

That specific change represents roughly a 30% throughput improvement.

For a single endpoint sample from the local benchmark runner:

Endpoint	Framework	Req/Sec	Avg Latency	P99 Latency
JSON Response	Oxarion	255271	0.785 ms	1.140 ms

The full generated benchmark report is written to benchmarks/Results.md.

Benchmark notes

• Benchmarks are run locally on Bun using oha
• Measured runs use sustained timed execution rather than short burst tests
• Results are more useful for comparing framework overhead on identical handlers than for making universal performance claims
• Numbers may vary based on hardware, Bun version, OS scheduling, thermal state, and background apps

If you want to reproduce the same style of measurement, see:

• Testing and benchmarking

Docs

Start here if you want the full guide.

• Docs index
• Getting started
• Server options
• Routing
• Route Injector
• Dynamic routing
• Middleware
• Request and response
• Production deployment
• WebSocket
• API reference
• Testing and benchmarking

License

MIT