Mountain Landscapers — mtnlandscapers.com

A rebuild of the existing Duda site as plain HTML, CSS and vanilla JavaScript. No framework, no build tooling to install, no JavaScript required to read any page. Every URL, page title, meta description and paragraph of copy from the live site is preserved, so the rebuild keeps the rankings the old site earned.

What's here

build.py            the generator — run this after ANY content change
pages.py            page inventory: slugs, titles, meta descriptions, FAQs
content.json        page copy, extracted from the live site
patches.py          deliberate corrections to that copy (each one documented)
legal.py            privacy policy, terms, accessibility statement
site.config.json    business facts: phone, address, hours, socials
serve.py            local preview server
css/style.css       one stylesheet for the whole site (~16 KB)
js/main.js          nav toggle + dropdowns (~2 KB), nothing else
img/                every photo, converted to WebP at 1600px and 800px
fonts/              Poppins, self-hosted
screenshots/        desktop + mobile captures of the built pages
*.html              generated — do not hand-edit, your changes get overwritten

Editing the site

The HTML files are generated. Never edit them directly — build.py rewrites them. Edit the source instead:

To change	Edit	Then
Phone, address, hours, socials	site.config.json	python3 build.py
A page title or meta description	pages.py	python3 build.py
Words on a page	content.json (find the slug)	python3 build.py
Legal pages	legal.py	python3 build.py
Colors, spacing, layout	css/style.css	python3 build.py
Add a photo	drop in img/, reference it in content.json	python3 build.py

build.py also regenerates sitemap.xml, robots.txt, llms.txt and 404.html, and re-stamps the ?v= cache-busting hash on the CSS and JS so returning visitors always get the current version.

Business facts live in one place

Phone, address and hours come from site.config.json and are written into every page's footer and structured data by the generator. After any change, confirm the number is identical everywhere:

grep -c "(865) 280-4642" *.html

Every page must match — Google cross-checks this against the Google Business Profile, and a mismatch costs local rankings.

Preview before you publish

python3 serve.py

Then open http://localhost:8787. The preview server mimics Vercel's clean URLs, so /retaining-walls works exactly as it will in production.

The edit → preview → publish loop

Production deploys from main only. Never edit main directly.

git checkout -b edits
# make changes, run python3 build.py, check with python3 serve.py
git add -A && git commit -m "Update Saturday hours"
git push -u origin edits

Pushing the branch gives you a Vercel preview URL — a private copy of the site with the change applied. Merging that branch into main publishes it.

Heads up: this project has Vercel Deployment Protection enabled, so preview URLs require a Vercel login. You can review them; a client can't, unless you turn protection off for preview deployments or enable a protection bypass.

Publishing

Already connected: pushing to main deploys to https://mtnlandscapers.vercel.app (project salt-services/mtnlandscapers). vercel.json carries the build settings, clean URLs, security headers, the 301 for the retired tree-service URL, and long-lived caching for images and fonts. Builds take about 4 seconds.

The custom domain is not attached yet — mtnlandscapers.com still serves the old Duda site.

Two things to know before editing

The street address is deliberately schema-only. 109 Bruce Street lives in site.config.json as address.streetPrivate and is written only into the JSON-LD, so the machine-readable NAP matches the Google Business Profile while no street address appears on the page. The visible address everywhere is "Sevierville, TN 37862". Don't print streetPrivate in a template, and don't remove it from the schema — the GBP listing depends on that match.

Retired URLs need a redirect, not a deletion. /tree-removal-and-service was removed when that service was discontinued; it 301s to the home page via vercel.json. If another page is ever retired, add the redirect at the same time — a bare 404 throws away whatever authority the page had.

The contact form is a lazy-loaded Coraline embed, on /contact only. Its IDs live in site.config.json under coralineForm; the loader is at the bottom of js/main.js. The iframe is injected only when the visitor scrolls near it or taps the button, which keeps a 1,695px third-party iframe out of the initial page load. If the client's form ID ever changes, edit site.config.json and re-run build.py — nothing else needs touching. Every other page shows cta_card() from build.py instead: call button plus a link to /contact.

Build settings are pinned in vercel.json (framework: null, empty build and install commands, output directory .) because the Vercel project was originally set up for a Next.js app. Leave those keys in place or the deployment will try to run next build again and fail.

Before go-live

See CLIENT-HANDOFF.md for the full checklist. The blocking item:

1. Submit a test lead through the form and confirm it arrives in Coraline

Then: point the domain at Vercel, add the Search Console verification tag, and submit sitemap.xml in Google Search Console.