← The AI Hype Audit — all 314 verdicts

PARTLY Partly true: the 20-item 'fix my SEO, make no mistakes' mega-prompt is a legitimate checklist Claude Code can mostly ship inside a repo, but the three things that actually get a vibe-coded site indexed (Search Console, server-rendered HTML, inbound links) live outside it Facebook · Sep 2026

The claimCaption: 'Google isn't showing my vibecoded website please make no mistakes... #claudecode #vibecoding #cursor #grokbot #codex.' A person reads one prompt to camera: 'Hey Claude, Google is ignoring my vibe coded website. I need you to build a sitemap.xml, add robots.txt, remove noindex tags, add canonical tags, add meta titles, add meta descriptions, use one H1 per page, fix header hierarchy, add alt text to images, add schema markup, add internal links, fix the broken links, compress images, improve Core Web Vitals, fix mobile responsiveness, enforce HTTPS, clean up URL slugs, add an OG image, verify Search Console, build a backlink strategy, make no mistakes.' The implied promise: paste this and Google starts showing your site.

The checklist is real SEO; the framing is theater. Of the twenty things the prompt asks for, about fourteen are file edits Claude Code can do in a repo it can see: sitemap.xml, robots.txt, strip noindex, canonicals, titles, descriptions, one H1, heading order, alt text, JSON-LD schema, internal links, a dead-link sweep, slugs, an OG image, and image compression via a build step. The rest are not repo tasks. 'Verify Search Console' needs the owner logged into Google to prove ownership; the agent can write the verification file or DNS record but cannot click Verify. 'Enforce HTTPS' is done by the host: Netlify and Vercel serve HTTPS and redirect HTTP out of the box. 'Improve Core Web Vitals' is a measurement over hosting, fonts and scripts, not a line item. 'Build a backlink strategy' produces a document, not links. 'Make no mistakes' does nothing. Meanwhile the reasons vibe-coded sites do not show up are usually elsewhere: a default Vite/React build ships an empty HTML shell that Google renders on a delayed second pass; Netlify and Vercel preview URLs carry an X-Robots-Tag: noindex header, so the link the builder has been sharing was never indexable; the site was never submitted to Search Console; and a brand-new domain with zero inbound links sits at 'Discovered - currently not indexed' for weeks. Google's own docs say crawling 'can take anywhere from a few days to a few weeks.' The poster is selling engagement, not a product: no course, no comment hook, just a hashtag stack.

What holds up

  • Transcript read against the caption (the reel itself is behind Facebook's login wall, so the audit worked from the transcript relayed to the self-thread). The 20 items sort into five buckets: 14 in-repo edits, 2 host-level (HTTPS, Core Web Vitals), 1 that needs the owner's Google account (Search Console), 1 off-site (backlinks), and 1 no-op ('make no mistakes').
  • Netlify docs (Deploy overview, 'Search engine indexing'): 'Requests to Deploy Previews, unpublished production deploys, and old branch deploys will have an X-Robots-Tag: noindex header included in the response.' Vercel KB: 'Vercel adds an X-Robots-Tag: noindex HTTP response header to every Preview Deployment automatically' (omitted only when a custom domain is attached to a non-production branch). Both hosts also serve HTTPS and redirect HTTP by default, so 'enforce HTTPS' is already done for the typical vibe-coded deploy.
  • Google Search Central, 'Understand JavaScript SEO basics' (last updated Mar 4, 2026): Google processes JS apps in three phases, crawling, rendering, indexing; pages are queued for rendering 'unless a robots meta tag or header tells Google not to index the page'; and 'server-side or pre-rendering is still a great idea because it makes your website faster for users and crawlers, and not all bots can run JavaScript.' Google's 'Ask Google to recrawl' page: 'Crawling can take anywhere from a few days to a few weeks,' URL Inspection requires an owner or full user of the Search Console property, and requesting a crawl 'does not ensure instant or eventual inclusion.'
  • Claude Code SEO tooling as of Sep 2026: Anthropic's official Marketing plugin ships a /seo-audit skill (keyword research, on-page, technical checks); community skills exist too (SE Ranking's seo-skills marketplace, AgriciDaniel/claude-seo with technical-SEO and schema sub-skills). None of them can verify a Search Console property, change a host's TLS setup, or earn a backlink.
  • Poster is a small vibe-coding tips page with no product, course, community or comment-for-link hook in the caption; the sell is reach via the #claudecode #cursor #codex #grokbot hashtag stack.

What doesn’t

  • One prompt across 20 unrelated concerns is a recipe for regressions; each item is a separate diff you should read before it deploys.
  • Four of the items (Search Console verification, HTTPS, Core Web Vitals, backlinks) cannot be closed from inside the repo, and the prompt implies the AI can close all twenty.
  • The most common actual causes, a preview URL carrying noindex or a client-rendered React shell, are not on the list at all; the prompt fixes symptoms on a site Google may never have been allowed to read.
  • 'Make no mistakes' is a superstition, not an instruction; nothing about the model's output changes because you typed it.

The catch

Run this prompt on a fresh React/Vite site and you will get a tidy pile of correct-looking edits, a robots.txt, a sitemap, JSON-LD in every head, and Google will keep ignoring you, because the site is still an empty HTML shell on a netlify.app or vercel.app URL that ships noindex, nobody verified it in Search Console, and no page on the internet links to it. The checklist is fine; it is the third step, not the first.

How to actually do it

  • Put the site on its real domain (not the .netlify.app or .vercel.app preview link), open Google Search Console yourself, verify ownership with a DNS TXT record, submit /sitemap.xml, and run URL Inspection on the homepage. The reason it reports (noindex header, 'Discovered - currently not indexed', blocked resources) is your actual to-do list.
  • If the build is a Vite/React SPA, have Claude Code switch to prerendering or a framework with static or server rendering (Astro, Next.js, or a prerender plugin) so the HTML response already contains the page text; confirm with curl or view-source, not the browser.
  • Only then hand Claude Code the on-page list, one item per prompt, in the repo: sitemap, robots, canonicals, titles and descriptions, one H1 and heading order, alt text, JSON-LD, internal links, slugs, OG image, image compression. Review each diff, redeploy, and re-run URL Inspection.
  • Earn a handful of real links (your social profiles, niche directories, a launch post, partners and clients) and wait; Google says crawling alone takes days to weeks and requesting indexing does not guarantee inclusion.

Most of the twenty items are real technical SEO and Claude Code can do them in a repo it has access to. Verifying Search Console, enforcing HTTPS, moving Core Web Vitals and earning backlinks are not repo edits, and the usual reasons a vibe-coded site is invisible (noindex on preview URLs, client-side-rendered shells, no Search Console submission, no links) are not on the list. Verified against Netlify and Vercel docs and Google Search Central; the reel itself is login-walled.

Confidence
High
Posted by
a 'vibe coding & AI' tips page (this reel: ~34K views, 228 reactions); reel sent to the self-thread Fri Sep 18, 2026, 7:59 AM

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.

© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.