kapps.io — static site

🌐 https://kapps.io

The kapps.io marketing site, served as flat files from GitHub Pages instead of WordPress on AWS.

WordPress remains the authoring tool. It is no longer the thing serving traffic: the pipeline in tools/ crawls the live WP install, strips the server-dependent parts, and emits a self-contained static tree into docs/.

Layout

Path	What it is
docs/	The published site. GitHub Pages serves this folder. Generated — don't hand-edit.
build/	Working output of the pipeline. docs/ is an rsync of it.
build.raw/	Untouched mirror snapshot, before transforms. Lets you re-run transforms without re-crawling.
assets/kapps-forms.js	Hand-written. The static form handler that replaced Contact Form 7.
tools/	The pipeline (see below).

Rebuilding

./tools/publish.sh              # full rebuild from the live WordPress
./tools/publish.sh --no-fetch   # re-run transforms against the existing snapshot

It never commits. Review the diff, then commit docs/ yourself.

Pipeline

Step	Script	Does
1	mirror.py	Crawls the site. Discovers assets by sweeping raw HTML/CSS with a regex rather than parsing the DOM, so it also catches lazy-load data- attributes and Elementor's JSON data-settings.
2	coverage.py + fetch_missing.py	Diffs the Yoast sitemaps against what was crawled and fetches orphan pages — published content no navigation links to. Found 23 on the first run.
3	transform.py	Removes CF7/reCAPTCHA/Akismet/rating/view-counter plumbing, strips WP head cruft, rewrites URLs to root-relative, remaps dead links.
4	repair.py	Re-fetches assets the crawler missed and converts RSS feeds to .xml.
5	finalize.py	CNAME, .nojekyll, 404.html, robots.txt, sitemap.xml, redirect stubs.
6	validate.py	Asserts every internal reference resolves to a real file and nothing still points at the WP host.
—	compare.py	Diffs the rendered text of every built page against live. Catches silent content loss a link-checker can't see.

What changed vs. WordPress

Forms. Contact Form 7 needed PHP. Replaced by assets/kapps-forms.js, which POSTs to a hosted form endpoint. reCAPTCHA is gone (it verified server-side); spam control is now a honeypot plus the provider's filtering. The old CF7 → gtag conversion event is re-fired on success, so form analytics keep working.

Set the endpoint in assets/kapps-forms.js:

var CONFIG = { provider: "web3forms", key: "<public-access-key>" };

Dropped. Post view counters and star ratings (both were WP AJAX). The static results display is kept where it existed; only the interactive vote widget is gone.

Redirects. Pages can't serve 301s. /join is a <meta refresh> + canonical stub.

Bugs fixed in passing

These were broken on the live WordPress site and are fixed here:

• ?page_id=28/30/49/51 — four service/technology cards linking to 404s
• /vr-ar, /3d-modeling-services, /game-development-outsourcing — dead footer links
• url(../wp-content/…) backgrounds on /portfolio/<item>/ pages — resolved one directory too shallow, so the block backgrounds 404'd on every deep portfolio page
• Junk slugs (cool_timeline/sadasd, 2021-2, …) removed from the sitemap

Known remaining 404s

Eight plugin CSS sprites and legacy IE webfont formats referenced by bdt-uikit.css / gwts-style.css. They 404 on the origin too — the files were never uploaded. Invisible in practice.

Hosting notes

• Pages must serve from /docs on the default branch.
• The repo must be public — the StansAssets org is on the free plan.
• Repo is ~60 MB against a 1 GB limit; bandwidth 100 GB/month soft limit.