regen-seo-audit
Run a technical SEO audit on regencivics.earth or a specific page. Covers meta tags, OG/Twitter card images, structured data (Event, Article, Organization, BreadcrumbList), sitemap and robots, interna
SKILL.md.claude/skills/regen-seo-audit/SKILL.md
GitHubReGen Civics Technical SEO Audit
What this skill does
Audit one page or the whole site against a tactical SEO checklist tuned to the actual constraints of regencivics.earth: a Vite + Wouter SPA with no SSR, hosted on Railway, content driven by tRPC. Spit out a prioritized fix list so Rye (or Claude Code) can ship the changes.
This is NOT a generic SEO skill. It assumes the site, the stack, the voice rules, and the FIXES_TO_MAKE doc convention.
Stack constraints to understand first
- Vite SPA, no SSR. Initial HTML payload is whatever's in
client/index.html. Search engines index the rendered page, but the more meta is in the static HTML the better for crawl efficiency, social preview, and bots that don't run JS. - Wouter for routing. Per-route meta has to be set client-side via a
helper (e.g.
<Helmet>fromreact-helmet-asyncif installed, or by directly mutatingdocument.headin auseEffect). Checkclient/src/lib/seo.tsor similar before rolling new patterns. - R2 for images. OG images live at
assets.regencivics.earth/..., proxied through/api/imgfor resize. Static OG images can sit at the R2 root path; dynamic OG (per quest, per event) goes through the proxy. - Public routes only. Anything behind auth (forum, profile, admin) gets
noindex, nofollow. Don't waste crawl budget on private surfaces.
The audit checklist
Run through these in order. Capture findings into a table with: Issue, Page(s), Severity (Critical / High / Medium / Low), Recommended fix.
1. Per-page meta (every public route)
-
<title>set, unique, under 60 chars, ends with " | ReGen Civics" -
<meta name="description">set, 140-160 chars, contains the page's core promise + a CTA verb (apply, join, explore, contribute) -
<link rel="canonical">set to the absolute https URL of the page -
<meta name="robots">if it should not be indexed (auth pages, draft pages, the dashboard)
2. Open Graph + Twitter cards
-
og:title,og:description,og:url,og:type,og:imageall set -
og:imageis at least 1200x630, hosted on R2 or CDN, absolute URL -
twitter:card=summary_large_image -
twitter:title,twitter:description,twitter:imageset (can mirror og: but Twitter's parser sometimes prefers explicit twitter: tags) - OG image actually shows the page's content, not a generic logo. Quests should have quest-specific images. Events should have event posters. Land projects should have project cards.
3. Structured data (JSON-LD)
Pick one or more types per page and inject as <script type="application/ld+json">:
| Page | Schema type(s) |
|---|---|
/ |
Organization + WebSite (with SearchAction) |
/events/:slug |
Event (with location, startDate, offers) |
/projects/:slug |
Article or LocalBusiness per project type |
/quests/:id |
Quest-style HowTo or Article |
/blog/:slug |
Article with author, datePublished |
/jobs/:slug (if any) |
JobPosting |
| Any list page | BreadcrumbList |
/economy, /governance |
Article with mainEntity |
Validate each block at search.google.com/test/rich-results before claiming done. Don't generate JSON-LD that Google's validator rejects.
4. Sitemap + robots
-
sitemap.xmlexists at the site root, lists every public route withlastmod, generated at build time (script inscripts/) or server-rendered through an Express route -
robots.txtallows public, disallows/forum/draft,/admin,/api,/dashboard,/profile -
robots.txtreferences the sitemap withSitemap: https://regencivics.earth/sitemap.xml - Submit sitemap once via Search Console (manual step, captured in handoff doc)
5. Internal link health
- Every public page is reachable from the homepage in 3 clicks max
- Critical pages (
/apply,/economy,/governance,/projects,/quests) appear in the global nav OR the footer - No orphaned routes (live in
client/src/App.tsxbut unreachable) - No broken internal links (
/path-that-404s). Check theuseLocationtarget of every<Link>against the App.tsx route table
6. Headings + landmarks
- One
<h1>per page, descriptive, contains the page's primary phrase - Heading order (h1 then h2 then h3) is hierarchical, no skips
- Page has a single
<main>element -
<nav>and<footer>landmarks exist on every page - Decorative images have
alt="". Content images have descriptive alt (50-125 chars, no "image of ...")
7. Performance signals (light pass)
For deep performance audits, hand off to the core-web-vitals skill. The
SEO-relevant checks here:
- LCP element on each page is identifiable and not blocked by a layout shift
- No render-blocking scripts in
<head>beyond Vite's required ones - Images use
loading="lazy"below the fold - Preload hints for hero images on landing pages
8. Mobile + accessibility (SEO-adjacent)
- Viewport meta tag set in
client/index.html - Tap targets at least 44x44 px on mobile (already audited in
regen-ship-gate, but recheck for landing pages) - Color contrast on all text passes WCAG AA (use
regen-form-designchecklist or web-quality-audit skill for the deep dive)
9. Voice rules (SEO copy)
Meta titles and descriptions must follow the project Writing Rules. Most common slip:
- Don't pack keywords. "Regenerative Civics | Regen Movement | Land Projects | ReGen Civics" reads as spam. Write a sentence a human would say.
- Em-dashes banned. Zero. Use a colon or period.
- No contrast framing in meta descriptions. "Not just a fund. A movement." is banned. State what it IS.
Output format: the audit report
When run as a full audit, produce docs/planning/FIXES_TO_MAKE_YYYY-MM-DD_seo.md
(use the regen-fixes-handoff skill for the doc structure). Tier the
findings:
- Critical: missing canonical URLs, missing OG images, broken JSON-LD that fails the validator, robots.txt blocking the homepage
- High: generic OG image on a high-traffic page, missing per-route meta, no sitemap, h1 missing
- Medium: thin meta descriptions, internal links to deprecated routes, duplicate titles across routes
- Low: alt text improvements, lazy-load gaps, breadcrumb schema missing on list pages
For each row include the Evidence column (file:line, screenshot, validator URL output) per the ship-gate convention.
When run on a single page
If Rye says "audit the SEO on /economy" or similar, run only the relevant checks against that one route, return a focused report (5-15 line list), and offer to ship the easy fixes inline if scope permits.
Quick wins to surface first
If asked "what's the highest-leverage SEO move right now," default to:
- Add a per-route
useSeo()helper if it doesn't exist, that sets title/description/OG/canonical from a single config object per page - Generate dynamic OG images for quests, events, and land projects via
/api/og?type=quest&id=...returning a 1200x630 PNG (or a static set committed to R2 if dynamic generation is too much scope) - Ship a
sitemap.xmlroute at the Express layer, queried from the tRPC routes that already enumerate public events / projects / quests - Add JSON-LD to the four highest-traffic pages: home, /apply, /economy, /governance
Cross-references
- Use
web-quality-auditfor the full Lighthouse + a11y pass - Use
core-web-vitalsfor performance deep-dive - Use
regen-fixes-handofffor the audit report doc - Use
regen-ship-gatebefore claiming any fix VERIFIED