Dev.to WebDev 🛠 Dev 👁 0 📖 4 min read

I audited 286 developer portfolios: 1 in 3 is a blank page to ChatGPT

Short answer: I ran 300 random developer portfolios through the same checks AI crawlers depend on. Of the 286 that loaded, 34% show fewer than 20 words to a bot that doesn't run JavaScript, which is how GPTBot and Claude

I audited 286 developer portfolios: 1 in 3 is a blank page to ChatGPT

Short answer: I ran 300 random developer portfolios through the same checks AI crawlers depend on. Of the 286 that loaded, 34% show fewer than 20 words to a bot that doesn't run JavaScript, which is how GPTBot and ClaudeBot read the web. Only 15% have an llms.txt, and 7% block AI crawlers outright, usually without the owner knowing.

People increasingly ask ChatGPT, Claude or Perplexity things like "find me a React developer who has built e-commerce sites". If those tools can't read your portfolio, you're not in the answer.

Bar chart: what AI crawlers find on 286 developer portfolios

How I measured it

I took a reproducible random sample of 300 sites from developer-portfolios, the biggest public list of portfolios on GitHub (2,100+ entries). Each one went through site-audit, a free tool I built that checks SEO, security headers and AI readiness.

The key detail: the audit reads the raw HTML the server returns, without running JavaScript. That's what AI crawlers see. 11 sites timed out or were down, and 3 returned an error, which leaves 286 live portfolios.

Finding 1: a third of portfolios are blank without JavaScript

98 portfolios (34%) contain fewer than 20 visible words in their HTML. Most are single-page React, Vue or Vite apps that render everything in the browser. To a human they look great. To a crawler that doesn't execute JavaScript, they look like this:

<body>
  <div id="root"></div>
  <script type="module" src="/assets/index-4f3a.js"></script>
</body>

No name, no skills, no projects. Google renders JavaScript, so these sites may still rank there. But Vercel's analysis of AI crawler traffic found that OpenAI's and Anthropic's crawlers fetch JavaScript files without executing them.

Widen the threshold to 250 words and it's 54% of portfolios. 35% have no <h1> in the HTML at all.

Fix: pre-render. Use static generation (Astro, Next.js output: 'export', Nuxt generate, SvelteKit prerender) or a pre-render plugin for your Vite app. A portfolio is static content. It doesn't need to be built in the visitor's browser.

Finding 2: 7% block AI crawlers, and most don't know it

19 portfolios block GPTBot, ClaudeBot and Google-Extended in robots.txt. When I opened those files, 17 of the 19 were written by Cloudflare, not by the owner. Cloudflare's managed robots.txt adds the blocks for you when the AI crawler setting is on.

If you put your portfolio on Cloudflare and want AI assistants to find you, open yourdomain.com/robots.txt and check. To allow them explicitly:

User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

Sitemap: https://yourdomain.com/sitemap.xml

Finding 3: almost nobody gives AI the context it needs

AI assistants prefer content they can quote with confidence: who wrote it, when, and what question it answers. Portfolios mostly skip this:

  • 91% have no publish or update date anywhere (meta tag, JSON-LD or <time>).
  • 88% have no question-and-answer structure (headings phrased as questions, or FAQ schema).
  • 85% have no llms.txt.
  • 63% have no structured data (JSON-LD) at all.
  • 49% don't declare an author.

The cheapest fix for most of these is one block of JSON-LD in your <head>:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Person",
  "name": "Your Name",
  "jobTitle": "Frontend Developer",
  "url": "https://yourdomain.com",
  "sameAs": ["https://github.com/you", "https://linkedin.com/in/you"],
  "knowsAbout": ["React", "TypeScript", "Accessibility"]
}
</script>

And an llms.txt at the root of your site, a plain Markdown summary written for language models:

# Your Name

> Frontend developer in Lisbon. I build fast, accessible React apps for startups.

## Projects
- [Project A](https://yourdomain.com/projects/a): e-commerce checkout, 40% faster load
- [Project B](https://yourdomain.com/projects/b): open-source date picker, 2k stars

## Contact
- Email: [email protected]

llms.txt is a proposal, and no major AI company has confirmed it reads it. It takes five minutes, though, and it doubles as a clean summary of your work.

The classic SEO basics are missing too

These aren't AI-specific, but they decide what shows up when someone shares your link:

  • 44% have incomplete Open Graph tags, so a link on LinkedIn, Slack or X shows no image or the wrong text.
  • 58% have a meta description that is missing (17%) or outside the 50–160 character range.
  • 57% have no sitemap.

The median portfolio failed 14 of 33 checks.

A 15-minute checklist for your portfolio

  1. View source (Ctrl+U). If you can't read your name and projects there, pre-render.
  2. Open /robots.txt and make sure you aren't blocking crawlers you want.
  3. Add one <h1> with your name and role.
  4. Add Person JSON-LD and a meta author.
  5. Add og:title, og:description and og:image.
  6. Write an llms.txt.
  7. Add a sitemap.

Or run your portfolio through the audit. It's free, needs no sign-up and checks all of the above in about ten seconds, with the exact fix for each item.

I don't name any individual site: the point is the pattern, not the people.

📰 Read the original article on Dev.to WebDev

Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.