# Make a Vue or Nuxt site readable to AI crawlers

> Vue SPAs and Nuxt with ssr false ship an empty shell to crawlers. Here is how to check it and the fastest way to fix it without a rewrite.

Canonical: https://pagerender.io/ai-crawlers/vue-nuxt  
Updated: 2026-11-11

Nuxt renders on the server unless you turn that off. A plain Vue app from Vite does not. The empty page ChatGPT sees is the second kind, and any Nuxt route you set to ssr false. If the source of a URL already contains the paragraph, that URL is fine.

[Check what AI crawlers see](https://pagerender.io/crawler-check/)

## Nuxt already renders, until a route opts out

Nuxt's default is server-side rendering. The document Nuxt sends includes the text of the page. A crawler that does not run JavaScript can read it. You do not need a prerender service for a route that already came from the server.

ssr: false in nuxt.config, or a route rule that sets ssr to false, switches that route to a client-rendered shell. People still get the app after hydration. GPTBot, ClaudeBot and PerplexityBot, which Vercel measured as not executing JavaScript, get the shell. The same is true of a Vite Vue project that never used Nuxt. Vue's own SSR guide treats server rendering as something you add, not the default of a Vite starter.

## How to see which mode you are in

View source on a product URL. If the title, the headings and the paragraph are in the HTML, the route is rendering on the server or was prerendered. If you see an empty app div and a script, it is a client route. The crawler check reports the word count of that source and a per-bot robots.txt line.

Do this per template. A marketing page can be server-rendered while the app behind /account is a client route on purpose. /account should often be noindex anyway. The pages you want cited are the ones to test.

## Fixes that match the mode

If the route was set to ssr false by habit, turn SSR back on for the public routes. That is a config change, not a rewrite, and it is the fastest fix when the page does not depend on window or a browser-only library during setup. Move browser-only code into onMounted so the server pass can finish.

If the project is a Vite Vue SPA with no Nuxt, you either add a server renderer, generate static HTML for the public routes, or put a prerender service in front of crawler user agents. The service is the one that does not restack the app. Pagerender serves the finished HTML to crawlers and leaves the Vue app for everyone else. A week on your own URL is enough to see the word count move.

Google still calls dynamic rendering a workaround and points at server rendering, static rendering or hydration. Use the workaround when the Vite app is the product and a Nuxt migration is a separate project. Use Nuxt's own SSR when you are already on Nuxt and a flag turned it off.

## Client-only libraries and route rules

A Nuxt page that imports a chart, a map, or an editor that touches window during setup will fail the server pass or render nothing until the client. The usual fix is a client-only wrapper around that widget, with the text of the page still rendered on the server. The widget can be empty in the HTML. The article next to it should not be.

routeRules in Nuxt can set ssr false for a pattern. A rule you added for the app shell, or a copy of someone else's config, is enough to make every matching URL a client page. Read the config before you add a vendor. One pattern of ssr false on a marketing route is the whole bug.

After the change, fetch the URL with JavaScript off. The crawler check does that and also shows whether robots.txt blocks the bot you care about. A server-rendered page that disallows GPTBot is still invisible to that bot. Rendering and permission are separate rows.

useAsyncData and useFetch in Nuxt run on the server and put the result in the HTML. A fetch you start in onMounted does not. The page can look identical in the browser and opposite in the source. When you move a call from onMounted to useAsyncData, view source again. The list should be in the document, not only in a JSON payload you hope a model will parse. Vercel noted that JSON already in the HTML can be read. A fetch that has not happened yet is not JSON in the HTML.

Hydration can then attach the Vue app to that HTML. A mismatch, where the server text and the client text differ, is a bug to fix, and it is a better bug than an empty document. The crawler already left with the server text. People get the client. Keep those two copies the same for the words you want cited.

A Vite Vue app that is not Nuxt has none of those helpers. There is no useAsyncData. The component mounts, then it fetches. The document a crawler saves is the mount point. You can add vite-ssg or move the public pages into Nuxt, and both are real projects. The smaller change, if the app is staying a SPA, is to render for crawlers at the edge and keep the Vite build for people. That is the prerender path. It does not make the app server-rendered for humans. It makes the crawler's copy match the page a person sees after JavaScript.

Check more than one route. A landing page you wrote as HTML inside the Vite index can pass while every route behind the router fails, because those routes do not exist as documents until the app boots. Request the route URL directly. If the server returns index.html for every path, which is the usual SPA fallback, the file is the same shell each time. The checker on /pricing and on / will then show the same thin document, which is the bug. Nuxt does not do that fallback for a server-rendered route: the URL is a document of its own. If your Nuxt URL behaves like the Vite fallback, SSR is off for that pattern. Turn it back on for the public routes, move browser-only setup into onMounted, and fetch the URL again before you pay anyone to prerender a page Nuxt would have rendered for free.

## Questions

### Does a Nuxt app with the default config need prerendering?

Usually no. Default Nuxt server-renders the page, so the HTML already has the text. Check a URL before you add a service. Prerendering matters for routes you set to ssr false, and for Vue apps that are not Nuxt. A passing check on a server-rendered route means the crawler already has the words.

### What about Nuxt generate?

nuxt generate writes HTML at build time for the routes it can resolve. Those files are readable without JavaScript. A route that only loads its data in the browser after generate still looks empty. Open the generated file for that route and read it. If the list is missing there, it will be missing for GPTBot.

[Try it free for a week](https://app.pagerender.io/signup?source=ai-crawlers-vue)