Static site for mvtech.solutions, published from main by GitHub Pages. Every page is generated from content/ by a small Node build so that metadata, structured data, breadcrumbs, dates and the AI-readable text files never drift apart.
| What | Where |
|---|---|
| Company facts, navigation, footer, timezone windows, "worked with" names | content/site.json |
| Home, About, Services overview, Solutions, Expertise, Process, FAQ, Privacy, Contact | content/pages/*.html — a <!--meta {…}--> JSON header (route, title, description, social-image text) followed by <main>…</main> |
Service and engagement pages (/services/*, /remote-consulting/) |
content/consulting.json |
| Insights articles | content/insights/<slug>.html — meta header (title, description, published date, keywords) followed by the article body. The filename becomes the URL. |
| Styles | css/input.css (design tokens at the top; .band-light switches a section to the light palette) |
| Behaviour (menu, sticky CTA, Calendly embed, email brief, analytics events) | js/site.js |
| Icons | js/icons-entry.js — import a Lucide icon before using a new data-lucide name |
| Factual public-site index for retrieval tools | llms.txt (hand-maintained; llms-full.txt is generated) |
FAQ blocks are plain <details><summary> markup; the build turns them into FAQPage structured data automatically and the tests fail if the two ever disagree.
Three names, each with one job, all set in content/site.json:
| Field | Value | Used for |
|---|---|---|
name |
MV.tech Solutions | The company name, written in full everywhere the company is named: copy, page titles, footer, copyright, og:site_name, structured-data name |
shortName |
MV.tech | Only what the logo artwork reads. It is an alternate name in structured data and the home-screen label in the web manifest, never a name for running text |
legalEntity.name |
BB & KV Inc. | The New York corporation behind the name, shown only where a legal party is needed: structured-data legalName, the About company information, the FAQ, the Privacy Policy, the Terms of Use and llms.txt |
MV.tech Solutions is a registered assumed name (DBA), not a separate corporation, so it never carries Inc. or LLC. npm test fails if a page says otherwise, if a page uses the short form as the company name (only /brand/ and /faq/ may print it, to explain the logo), or if the legal entity appears outside the pages listed above. The legal entity is spelled in content/site.json only. Page sources use tokens instead of retyping it: {{legal.name}}, {{legal.form}}, {{legal.line}} (the full sentence), {{legal.party}} (the contracting-party form), {{company.name}} and {{company.short}}. The build fills them and fails on an unknown token. llms.txt is hand-maintained, so npm test checks that it states the same legal line.
The company has two offices, always listed New York first (offices in content/site.json). Each office has streetAddress, locality, region, postalCode, country, countryCode, telephone and telephoneDisplay. Those fields feed the footer address row, the <!--offices--> block (on the home, About and Contact pages, with heading and intro from officesBlock), the social images, and the structured-data address, location and contactPoint. A street address, ZIP code, phone number and "Get directions" link appear as soon as the fields are filled; until then the office shows city, state and country only. Update the Offices line in llms.txt at the same time. Staff work from both offices. The India GSTIN belongs to the India operation and is published as a labelled identifier, never as the US entity's tax ID.
/privacy/ (Privacy Policy), /terms/ (Terms of Use) and /accessibility/ (Accessibility statement) are linked from every footer through legalLinks in content/site.json, and npm test fails if one is missing. Each prints a "Last updated" date from content/page-dates.json through the <!--last-updated--> marker, so the date moves only when the text changes. The Privacy Policy states that Global Privacy Control and Do Not Track signals turn analytics off, and that Google signals and ad personalization are disabled; the analytics loader in scripts/build.mjs does both, and npm test checks it. Change the code and the policy together. The site is written in US English.
GitHub Pages cannot send server redirects. A route listed in redirects in content/site.json (for example /us-clients/ → /about/) is written as a small noindex page with an instant refresh and a canonical to the target, which search engines treat as a permanent move. It stays out of the sitemap and llms-full.txt.
The name, the legal entity and the offices went live together on 14 November 2026. When any of them changes again, update content/site.json, the pages listed above and llms.txt in one release, then work through the listings checklist in docs/seo-maintenance.md so the same facts appear everywhere search and AI engines read them.
downloads/MVtech_Brand_Essentials.zipis the brand kit download on/brand/. It is a copy ofMVtech_Brand_Essentials_v3.zipfrom the local brand kit, which git ignores.brand/email/holds the email-signature images: a copy of the brand kit's05_Email_Signature/images/. The kit's-hostedsignatures load them fromhttps://mvtech.solutions/brand/email/. Keep the file names: signatures already pasted into mail clients point at them.npm testfails if one of the nine images the signatures use goes missing.
npm ci
npm run build # pages, sitemap.xml, llms-full.txt, 404.html, css/site.css, js/icons.js
npm run build:og # social-share images for new or changed pages (needs Chrome/Chromium installed)
npm test # metadata, structured data, links, sitemap, registry, robots, manifest
git add -A && git commit -m "…" && git pushCommit the generated output together with the source change; GitHub Pages serves the committed files and the Validate website workflow rebuilds and fails if anything was forgotten. The Notify search engines workflow waits for this commit's Pages deployment, then submits changed URLs to IndexNow. Submission helps participating engines discover updates; it does not guarantee indexing or AI citations. Failed submissions fail the workflow so they remain visible.
content/page-dates.json records a content hash and the published/modified dates for every page. The build updates a page's modified date only when its content actually changes, so <lastmod> in the sitemap and dateModified in structured data stay truthful. Do not edit it by hand.
npm run build:og renders images/og/<slug>.jpg for each page listed in content/og-manifest.json. It uses an installed Google Chrome, Edge or Chromium; set PW_EXECUTABLE=/path/to/chrome if none is found, and OG_FORCE=1 to regenerate everything after a design change.
- Create
content/pages/<name>.html(with aroute) orcontent/insights/<slug>.html. npm run build && npm run build:og && npm test.- Link to it from a related page and, for a top-level page, from
nav/footerincontent/site.jsonand fromllms.txt.
GA4 (G-YRM2SJ2EMP) runs only on the public website hosts, keeping local previews out of production traffic. It records contact_intent for email, phone and booking link clicks (method, service, page_path), book_call when the actual embedded Calendly iframe confirms a booking, and brief_compose when a visitor builds an email brief. Booking buttons lead to /contact/#book-call; an external Calendly fallback remains on the contact page. Only book_call confirms a booking; register it as a key event in GA4. External bookings and received emails require separate reconciliation.
Service pages declare three relevant relatedServiceKeys in content/consulting.json; articles declare relatedArticleRoutes in their meta header. These links are validated during the build so discovery follows the topic rather than file order. npm test also checks production-only analytics, contact intent, booking message validation and calendar load failures.
See docs/seo-maintenance.md for the search and AI-visibility checklist.