[maidulcu]

What's new

A plain-HTML features page (features.html) that search engines and AI assistants can read. The game itself is a 3D canvas with almost no text for them.

• SEO:
  • Every feature with a screenshot and a plain description.
  • "Why MiniCrafter is different", the two editions, safety and privacy, and a section for parents and teachers.
  • A good title, description and social-sharing tags.
  • sitemap.xml and robots.txt.
• AEO: a 10-question FAQ with short, direct answers ("Is it free?", "Is it safe for kids?", "Does it work on iPad?", "How is it different from Minecraft?"…), marked up as FAQPage.
• GEO:
  • Factual, quotable sentences, and a VideoGame structured-data block (free, platforms, ages, players, publisher).
  • An llms.txt summary for AI assistants.
• New screenshots: Course Maker, Photo Mode and Play Together.
• Links from the game menu ("✨ All features") and the README.

Works on any domain

• Nothing site-specific in the page: no domain is hard-coded.
• tools/deploy-site.sh now builds the site in a temporary folder, so your checkout is never changed. It adds the site-specific parts at deploy time: the canonical address and social tags on features.html, sitemap.xml and robots.txt. Renaming the domain is just a change to SITE_URL in tools/.deploy.env.

Launch checklist

The online site still tells search engines not to index it (beta). At launch, remove two things in the server's nginx config, so the generated robots.txt and sitemap take over:

• the pre-launch robots.txt override;
• the X-Robots-Tag: noindex header.

Tests

• Structured data: both blocks are valid JSON, and the FAQ data matches the questions shown on the page.
• In the browser: 12 feature cards, no broken images, and no sideways scrolling at 390 px (phone) or desktop.
• Deployed to the online edition: the page, llms.txt, sitemap.xml and the screenshots load, and the canonical tag has the site's address.
• Smoke test: full test passes locally.

🤖 Generated with Claude Code