sportsdata-site

Static marketing site for sportsdata-ai.com, served from GitHub Pages. Source of truth for the product pages lives in the private product repo; publish updates with its scripts/deploy-site.sh.

Layout

Path	What it is
index.html	Home page. Tool catalogue browser, recorded demo, FAQ, download.
engines.html	Generated — do not edit by hand. Run python3 scripts/build_engines_page.py.
board/, sports/	The two board terminals — LIVE against the homelab backends (live. / sportslive. subdomains via Cloudflare Tunnel); */data/replay.json is the automatic fallback when a backend is unreachable.
search.html	Site search — indexes the page list plus catalogue.json.
sitemap.html	Human-readable sitemap.
thanks.html	Post-download landing page with install steps. Linked from the download CTA.
404.html	Custom 404 (GitHub Pages serves this automatically).
maintenance.html	Standalone placeholder. Not linked; swap in manually if ever needed.
terms.html, privacy.html, accessibility.html	Legal and compliance pages.
assets/	Favicon set and the Open Graph share card.

SEO files

robots.txt, sitemap.xml, site.webmanifest and llms.txt live at the root. .nojekyll stops GitHub Pages running the files through Jekyll.

When you add or remove a page, update all four of these:

1. sitemap.xml — add the <url> entry and bump <lastmod>.
2. sitemap.html — add the human-facing row.
3. search.html — add an entry to the PAGES array so site search can find it.
4. llms.txt — add it under ## Pages.

Every indexable page carries a <link rel="canonical">, Open Graph and Twitter card tags, and JSON-LD (BreadcrumbList on subpages; WebSite / Organization / SoftwareApplication / FAQPage on the home page). Utility pages (404, search, thanks, maintenance) are noindex, follow rather than blocked in robots.txt, so crawlers can actually see the directive.

The home-page FAQ markup and the FAQPage JSON-LD must stay in sync — Google requires the answers to be visible on the page.

Regenerating assets

The share card and icons were rendered from SVG with macOS qlmanage + sips:

qlmanage -t -s 1200 -o . icon.svg && sips -Z 180 icon.svg.png --out assets/apple-touch-icon.png

assets/og-cover.jpg is 1200×630. Keep those dimensions — they're declared in the meta tags.

Support links

Ko-fi (ko-fi.com/danieltomaro) is linked from every page footer, the #get section, the thank-you page and .github/FUNDING.yml. Two rules the copy follows deliberately:

• The ask never reads as a condition of the download. On the home page it sits in its own note above the legal one, and says "both are free forever" first.
• terms.html §3 and privacy.html §1 state that a donation is a voluntary gift that unlocks nothing and gates nothing, and that Ko-fi handles payment so we never see card details. Keep those in sync if the funding setup changes.

Analytics & consent

Page analytics is Cloudflare Web Analytics — the beacon is at the bottom of every page and the token is a public site identifier, not a secret. It is cookie-free with no fingerprinting and no cross-site tracking, which is why the site has no cookie banner; privacy.html §1 says so explicitly. Add the beacon to any new page you create.

window.ANALYTICS_URL in index.html is a dormant, optional GoatCounter hook set to null. If you ever point it at a script that does set cookies, the consent gate makes it opt-in: nothing loads until the visitor accepts, and the choice is stored in localStorage rather than a cookie. Update privacy.html if you turn it on.

Provider and tool counts

Do not hand-type them in index.html. The counts come from catalogue.json at runtime via data-cat="providers", data-cat="tools" and data-cat="headline" spans — they were once hardcoded and sat at "29 providers · ~522 tools" long after the real figures had passed 60 and 750. Prose elsewhere (llms.txt, search.html, sitemap.html, 404.html, the JSON-LD and assets/og-cover.jpg) is static, so it says "60+" and "750+" rather than exact numbers.