Build page templates that crawlers and assistants read well
Builds a site’s page templates and shared components so crawlers and assistants read every page well, and proves each pattern with clickd. About structure and spacing, not color, type or brand.
Add it, from the top of the site’s repositorynpx skills add subinternal/clickd --skill clickd-page-system
Then ask your agentRework our page templates so every page has one h1, the right structured data and a breadcrumb, then check them with clickd.
It works on the templates, not page by page: one frame shared by every page, one outline, the structured data for its kind, and the parts that carry weight. The look stays yours.
A service page, as the skill builds it
Point at a part to find it on the page.
Sample
header, nav
breadcrumb
one h1, then h2
JSON-LD
ServiceproviderBreadcrumbList
table, image, date
links in the text
footer
tel:/llms.txt
The frame, built oncetitle, description, canonical, share-image, content-in-htmlStep 2
A breadcrumb below homebreadcrumb, within-three-clicksStep 5
One outlineone-h1, question-headings, answer-firstStep 3
Data for its kindstructured-data, schema-fitsStep 4
Parts that carry weighthas-table, image-alt, dated, sources-shownStep 6
Linked into hubslinked-in-context, no-orphans, direct-linksStep 5
A way to get in touchphone-link, next-step, llms-txtStep 2
First, name each kind of page
clickd gives each kind its own rules, so a page filed under the wrong kind is checked against too few. Where the address cannot change, the structured data says the kind.
HomeThe site's front pageOrganization, WebSite
Service or productOne offer: a service, a product, a location servedService, BreadcrumbList
ArticleA guide, a post, a how-toArticle, BreadcrumbList
HubA list of pages of one kind, such as /servicesCollectionPage
You are building a new site, or a new kind of page, and want it right from the first template.
clickd’s findings point at many pages made from one template.
You are reworking a layout, a component library or a design system and want its HTML sound.
Leave it when
You want a new look. This changes what is under the page and puts its spacing on one scale, never its color, type or brand.
The job is product data on a shop: the shop ready skill covers it field by field.
You want copy written. The patterns say where the answer goes; the words come from the site.
Install it by hand
The line at the top adds it for any agent. Or fetch its 12 files where your agent looks. Your agent also needs clickd connected over MCP: connecting an agent shows how.
Loads skills from .claude/skills/ in the repository, for everyone who works in it, or ~/.claude/skills/ for every project on your machine. Claude uses it when a request matches its description, or type /clickd-page-system.
Puts the files in .claude/skills/clickd-page-system/.
Loads skills from .agents/skills/ in the repository and its parent folders, or ~/.agents/skills/ for every project. Codex uses it when a task matches its description, or mention it as $clickd-page-system; /skills lists them.
Puts the files in .agents/skills/clickd-page-system/.
Loads skills from .cursor/skills/ or .agents/skills/ in the repository, or ~/.cursor/skills/ for every project. It also reads .claude/skills/, so one copy there serves both. The agent uses it when it is relevant, or type / in Agent chat and pick it.
Puts the files in .cursor/skills/clickd-page-system/.
An agent that does not load skills can still follow the file. Save the folder in the repository and point its standing instructions at it, with a line in AGENTS.md, which many agents read. The agent reads AGENTS.md when it starts, and follows the file when the task matches.
SKILL.md stays short and sends your agent to a reference when a step needs it.
SKILL.mdThe eight steps from naming each kind of page to setting its spacing and proving each pattern with clickd, and the table of patterns, rules and free checks.194 lines
references/page-kinds.mdA template for the home, service, product, article, hub and plain pages: the outline, the structured data, the links in and out, the rules each answers, how clickd files a page, and a service template before and after.204 lines
references/structured-data.mdJSON-LD for each kind of page, how to escape it, FAQ only where it is real, what never to mark up, a full article graph, Google’s required fields, building a block from data and what breaks it, and Google’s and schema.org’s pages for each type.246 lines
references/components.mdThe frame, breadcrumb, question and answer, table, image, byline, sources, contact block, footer and links in the text, each as the HTML it should send.173 lines
references/outline-and-accessibility.mdHow clickd reads each heading rule, an outline before and after, landmarks and the skip link, language, focus and accordions, and how to check each.102 lines
references/links-and-hubs.mdThe tree from home to hubs to pages, hub and spoke worked, links a crawler can follow, paged lists, town pages that read as doorways, and the sitemap.127 lines
references/images-and-speed.mdAlt text decided case by case, the image element before and after, the order of speed work, and the commands that check compression and first byte.92 lines
references/spacing.mdFinding the scale a site already uses, writing it as tokens for CSS or Tailwind, moving every padding, margin and gap onto it a component at a time, and a measure to run in the browser before and after.105 lines
references/spacing-traits.mdThe spacing rules, one to a section with an id that stays the same: twelve measured by code and four judged from screenshots, each with a bad and a good example.135 lines
scripts/measure.mjsMeasures a page against the twelve measured spacing traits at 1440 and 375 wide, in the Chrome already on the machine, with nothing to install. Exits 1 when a trait fails.317 lines
references/migration.mdMoving a live site onto the system a template at a time: the order of releases, what not to change, moving an address, and notes for Next.js, Shopify, WordPress and single page apps.87 lines
references/troubleshooting.mdA rule that still fails after the fix, by symptom, cause and fix, and the traps that look like success.49 lines
17Build and submit a sitemap, Google Search Central.Full canonical addresses, accurate lastmod, the size limits, and the robots.txt line.
18Spam policies for Google web search, Google Search Central.Doorway abuse and scaled content abuse, the risk in town pages that swap only a name.
19Understand the JavaScript SEO basics, Google Search Central.Rendering, why server rendering is still advised, and that not every bot runs script.
20How to move a site, Google Search Central.Mapping old addresses, permanent server-side redirects in one hop, updating links and sitemaps, and keeping redirects a year or more.
21Redirects and Google Search, Google Search Central.Which redirects Google reads as permanent and which as temporary, and why server-side is best.
schema.org
22LocalBusiness, schema.org.The type and its properties.
23Service, schema.org.The type for a service page.
24Article, schema.org.The type for an article page.
41Optimize Largest Contentful Paint, Google, on web.dev.Never lazy load the main image, high fetch priority, the image findable in the HTML, and time to first byte.