Generic server

Fetch-style frameworks

Use the response helpers with Hono, Bun, Cloudflare Workers, Elysia, or any framework that accepts a standard Response:

import {
  serveSW,
  servePreviewBridge,
  servePreviewBridgeScript,
} from '@scelar/nodepod/server';

app.get('/__sw__.js', () => serveSW());
app.get('/__nodepod_bridge__.html', (context) =>
  servePreviewBridge(context.req.raw));
app.get('/__nodepod_bridge__.js', () => servePreviewBridgeScript());

Express, Fastify, or node:http

Node-native helpers return a body and header map:

import { serveSWNode, servePreviewBridgeNode } from '@scelar/nodepod/server';

app.get('/__sw__.js', async (_request, response) => {
  const { body, headers } = await serveSWNode();
  for (const [name, value] of Object.entries(headers)) {
    response.setHeader(name, value);
  }
  response.status(200).send(body);
});

The bridge helpers return the response body and headers, so mount both paths using the same response-writing pattern. For the HTML bridge, forward a mode=top or mode=parent query when your preview host uses the top-level bootstrap handoff. The complete Express example is in Preview deployment.

Only route the documented Nodepod paths to these handlers. Your application remains responsible for authentication, caching policy outside those paths, and wildcard preview-host routing.