Switch from another prerender service in 5 minutes

If your site already sends crawlers to a prerender service, you keep your middleware and change two values: the service URL and the token. Prebaked speaks the same protocol as the open-source prerender middlewares (prerender-node, prerender_rails and the common nginx config), so there is no new code to write.

Find your plan

What changes and what stays

Before you start

1. Sign up for Prebaked.
2. Add and verify your domain (for example www.<your-domain>), with a DNS TXT record or a file. Prebaked renders only verified domains. Until the domain is verified, it answers 403.
3. Copy your token. It is shown once. You can rotate it later in the dashboard.

Step 1: change the service URL and token

Use these values everywhere:

Use https://. Prebaked refuses plain http:// calls with 403, because the token would travel in clear text, and it never redirects them.

prerender-node (Express, Connect)

Change the two settings wherever you set them now:

prerender
  .set('prerenderServiceUrl', 'https://service.prebaked.io/')
  .set('prerenderToken', process.env.PREBAKED_TOKEN);

Or set the environment variables the middleware reads:

PRERENDER_SERVICE_URL=https://service.prebaked.io/
PRERENDER_TOKEN=<your token>

Recommended while you are there:

• Add the afterRender hook, so crawlers get your normal page instead of an error if the prerender service fails.
• Add the AI crawlers that prerender-node 3.8.3 does not recognise:

prerender.addUserAgents(['OAI-SearchBot', 'ClaudeBot', 'Claude-User', 'meta-externalagent']);

Full steps: Express install guide.

prerender_rails (Rails, Rack)

Change the prerender_service_url: and prerender_token: options of Rack::Prerender, or the environment variables:

PRERENDER_SERVICE_URL=https://service.prebaked.io/
PRERENDER_TOKEN=<your token>

prerender_rails defaults to http://, so set the https:// URL explicitly.

nginx

In your current config, change the prerender host in proxy_pass (or the variable that holds it) to service.prebaked.io, use https://, and put your Prebaked token in proxy_set_header X-Prerender-Token.

The usual nginx prerender config passes every service error on to the crawler. Prebaked's config falls back to your site on errors and timeouts and includes Prebaked's crawler list, so consider replacing your block with it: nginx install guide.

Cloudflare Worker, Netlify, Vercel or another edge script

If your current service gave you a worker or edge script, change its service URL and token the same way. Or replace it with Prebaked's version, which falls back to your site on errors and after 5 seconds:

• Cloudflare Worker
• Netlify Edge Function
• Vercel or Next.js middleware

Step 2: deploy and check

curl -A "Googlebot" https://www.<your-domain>/        # rendered HTML from Prebaked
curl https://www.<your-domain>/                      # your normal page

The first crawler request appears in the request log in the Prebaked dashboard. The first request for each page renders it, and later requests are served from the cache. Cache hits are free: you pay only for fresh renders.

You can also run the free checker to compare what crawlers without JavaScript get with what people see.

See what ChatGPT sees on your site

What is different with Prebaked

• HTTPS only. Plain http:// service calls get 403.
• No _escaped_fragment_ URLs.
• Error codes. Wrong token: 401. Domain not verified: 403. With the recommended fallback setups, crawlers get your normal page instead of these errors.
• Never worse than today. Prebaked answers within its own time limit. If a page is not ready, it returns your origin page (fetched without JavaScript, with its own status code) and finishes rendering in the background for the next crawler. With the edge integrations, crawlers also get your site if Prebaked is unreachable or slower than 5 seconds.
• AI crawlers included. Prebaked's crawler list includes ClaudeBot and OAI-SearchBot.
• Pricing: you pay for fresh renders, and cache hits are free. Plans: pricing.

Switching from Rendertron

Rendertron is an open-source headless Chrome rendering service, published in Google's GoogleChrome organisation on GitHub. Its repository says the project is deprecated and will not be actively maintained, and it was archived on 6 October 2022.

Rendertron uses its own middleware (rendertron-middleware), which is configured with a proxyUrl pointing to your Rendertron instance and has no token setting. So the switch is not only a URL change. Replace the Rendertron middleware with one of the Prebaked setups:

1. Express: replace rendertron-middleware with prerender-node set to Prebaked's service URL and token. See the Express install guide.
2. Any other stack: put the Cloudflare Worker, Netlify, Vercel or nginx setup in front of your site.
3. Shut down your Rendertron instance after the check in step 2 passes.

You no longer run and update a headless Chrome server yourself.

When you may not need a prerender service

Google describes dynamic rendering (serving crawlers a rendered version of the page) as a workaround and not a recommended solution. If you can move your app to server-side rendering or static generation, crawlers get the content without any service in between. A prerender service fits when changing the app is not practical. Background on why AI crawlers need the content in the HTML: Can ChatGPT read JavaScript? and AI crawlers and JavaScript sites.

FAQ