skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
docs.clickmax.io89 installs

clickmax-pages

Use when the user wants to create, inspect, restyle, rebuild, clone, configure, or publish a native Clickmax-hosted page.

How do I install this agent skill?

npx skills add https://docs.clickmax.io --skill clickmax-pages
view source ↗

Is this agent skill safe to install?

No partner audit is available yet. Read the source before installing.

What does this agent skill do?

As tools abaixo aparecem com os nomes que o MCP da Clickmax registra. Se o seu cliente de IA prefixar nomes de tool (mcp__<servidor>__, mcp_<servidor>_, ou outro), use o nome já prefixado que aparecer na sua lista de tools.

When this applies

Any native Clickmax page job: discovery, inspection, authoring a page with real content, a targeted edit, a clone, configuration, checkout binding, or publication.

Scope boundary — pick the right skill first:

SituationSkill
Page already lives on a domain NOT connected to Clickmax (Framer, Webflow, own site)clickmax-external-pages
Whole connected funnel with several pages and edgesclickmax-funnels
Offer, price, guarantee, deliverableclickmax-offers
Form/quiz behavior, fields, logicclickmax-forms-quizzes
One native page — content, visual system, structure, publishthis skill

Key assumptions

  • A page with real content is authored as one full HTML document and imported. pages_create alone produces a blank page or a template copy — no copy, no sections, no working form.
  • The authoring path is validated by machine, not by taste: recipe capabilities and every requiredSections id are checked before import. Guessing the shape wastes a round trip.
  • Default design = a catalog design (page_designs_list → page_designs_get → slug in manifest designIdOrSlug). Its palette/type become the page tokens; the style guide only keeps logo + brand name. New page, no design named, restyle, or the user asks to see/list the designs → question type pageDesign (cards + full catalog in the chat), never a text list; skipped → choose the best fit and say which. Bundled editorial HTML = section structure only, or full fallback when nothing fits → visual system.
  • The project's style guide is a fresh default (commonly light mode, blue/navy palette, small type scale) until someone changes it, and no tool changes it. A dark/accent request that conflicts with it must be satisfied by an explicit local palette applied to every section — never by half-styled markup that lets bare tags inherit the default.
  • Imported author scripts are stripped. Use native HTML/CSS interactions or the documented declarative motion contract.
  • Page created from the Pages list AI box = placeholders: name Nova página|New page|Nueva página, type guessed from the goal chip, path /rascunho-<id>, project default domain. After authoring → pages_update name|type|description|slug path from the content; never change domainId (the user picks it in the first-publish review).
  • Images: real asset known → use it; otherwise mark a reserved image slot (data-cx-image-slot + hint + kind) from the context's imageSlots, never a guessed URL → components.
  • One capture form per page. Two forms record the lead twice and make the funnel's form_submit trigger ambiguous.
  • Missing facts/assets → content discovery before authoring. Explicitly requested fictional samples stay visibly labeled in draft; replace/remove before publication. Never invent actual payment, delivery, contact or access links.

Thought process

  1. Resolve the target page first — existing page, new page in a project, or a page inside a funnel. Never act on a remembered id without a fresh lookup this turn.
  2. Classify the job: read/config · blueprint only · new page with content · targeted edit · explicit full rebuild · clone · publish.
  3. Narrow request on an existing page → smallest possible edit. Never regenerate unrelated sections.
  4. Resolve account facts and ask the unanswered content discovery questions. A request for the default design does not skip discovery.
  5. Pick the recipe, read its bundled HTML base (or the explicit reference), then select composition variants and map verified content into slots. Reuse the funnel palette.
  6. Author, validate, fix every error, import as draft.
  7. Publication is always a separate, explicitly approved step (approval card, or the user's written "pode publicar" in the latest message).

Execute guide

  1. Content discovery — ask only missing facts/assets, including proof and real next-step URLs: content discovery.
  2. Pipeline and tool ordering — recipes, manifest, generation context, validation error codes, target rules, clone, configuration, and publication: authoring pipeline.
  3. Visual system — choosing the curated design, mapping it onto the injected tokens, type scale, color ramp, spacing, container, and breakpoints: visual system.
  4. Structure and copy — the mandatory spine, page jobs, section logic, and copy rules: sections and copy.
  5. Components — the CSS-only building blocks that survive import, plus the form, checkout, CTA, and motion contracts: components.
  6. Whop/Hotmart/Stripe/Pagar.me/PagBank checkout — selling through the workspace's own account: integration check FIRST (no integration → instruct the user to connect it, never build without it), catalog walk, externalCheckout, limits to tell the user: external checkout.

Report

  • Open with the page job, target product/offer, and any material assumption made.
  • Blueprint answers: sections in order as section → job → final copy → visual treatment; cap long lists with +N more sections.
  • Build/edit answers: the visible sections, the chosen design, the responsive behavior, and exactly what changed. Do not dump markup unless asked.
  • List missing business facts and assets as explicit gaps, named.
  • Uncertain or failed save → say so and say what must be re-checked. Never claim success from a call that did not confirm.
  • Before reporting a finished layout, inspect the imported draft on desktop and mobile; repair overflow, missing assets, typography and inactive actions. If no browser is available, disclose that visual review is pending. Publication remains a separate opt-in.
  • After a create/publish that finishes the requested work, follow clickmax-getting-started to close with at most one opt-in offer of the next setup task.

Warnings

  • Never publish without explicit consent, and never as part of a build step.
  • Never replace a non-empty page unless the user asked for a full rebuild.
  • Never hand-write checkout markup — the payment runtime looks for specific ids and a mistake breaks charging with no visible symptom.
  • Whop/Hotmart/Stripe/Pagar.me/PagBank checkout requested and the integration is missing or incomplete → instruct the user to connect/complete it in Integrações and stop; never deliver that page without its checkout.
  • Never invent an image URL. A broken image in production is worse than a section without one.
  • A validation call that returns valid: false means fix and revalidate — not try a different tool.
  • Warnings on a successful import are actionable: re-import the same page id with the corrected CSS.

Anti-patterns

  • Recreating the default visual system from prose instead of adapting the bundled HTML base.
  • A generic "modern and clean" direction with no concrete visual concept.
  • Writing copy before deciding the page job, the audience's awareness level, and the objection being answered.
  • The same centered card layout in every section.
  • Padding the page with schedule, modules, instructors, certificate, and bonus blocks the user never gave data for.
  • Regenerating the whole page for a color, spacing, or one-section request.
  • Creating a blank page and telling the user to fill it in themselves.

Clickmax skill revision: 61fff086819e

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/docs.clickmax.io/clickmax-pages">View clickmax-pages on skillZs</a>