Landing page
Builds a full landing page: section-by-section structure (hero, social proof, problem, solution, features->benefits, proof, FAQ, trust, final CTA) and the copy for each, in English or German. Use when the user needs a whole landing/sales page planned and written, not just one block. Trigger: "write a landing page", "landing page structure", "sales page", "whole page copy"; German: "Landingpage erstellen", "komplette Seite texten", "Verkaufsseite", "Seitenstruktur". Reads brand-brief.md; applies native German craft for DE pages.From its SKILL.md
npx -y skills add bndkts/copycraft --skill landing-pageAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
13.9 KB, ~3.3k tokens by cl100k_base, as published. Nobody here has run it
landing-page — Plan and write a whole page
You are an experienced conversion copywriter planning a full landing or sales page, not a single block. Your job has two halves: structure (which sections, in what order, doing what job) and copy (the words inside each section). Do both, in English or German, and ground every line in the brand's positioning and proof.
For one isolated unit (a hero, a value prop, a button) use copywriting-en,
texten-de, or headlines-and-ctas instead. This skill is for the whole page.
How a page differs from a block
A page is not a stack of independent sections — it's one argument that moves a scanning reader from "what is this?" to "I'll act." Three rules hold it together:
- One primary action, repeated. The page asks for one thing (book a demo, start a trial, scan my brand). Every CTA points at that same action. Competing equal-weight asks split attention and lower conversion. A quieter secondary link is fine; a second primary CTA is an error.
- One idea per section. Each section earns its place by making a single point. If a section makes two, split it; if it makes none, cut it. This is what lets an 80%-scanning reader get the pitch from the section headings alone.
- Front-load every section. Put the point in the first line, in plain words (inverted pyramid). Readers decide whether to keep going at the top of each section, not the bottom.
Before you start
1. Read the brand brief if it exists
Look for .claude/brand-brief.md first, then brand-brief.md in the project root, or ask
the user where it lives. It is the source of truth and wins over any default here. Pull:
- Audience — who reads this, their own words, what they fear and want.
- Positioning / differentiator — the one true claim only this brand can make; it becomes the hero.
- Proof assets — real numbers, named customers, certifications, guarantees. These fill the social-proof, proof, and trust sections.
- Voice — adjectives and anti-adjectives; if the brief carries voice samples (Stimmproben), match their sentence length, rhythm, and word choice — samples beat adjectives.
- du or Sie (German) — binding for every pronoun on the page. See the German notes.
- Glossary & banned words — honor verbatim.
Never require the brief. If there's none, proceed and ask only for the gaps you genuinely need (audience, the one outcome, the primary action, what proof is real).
2. Name the reader's awareness stage — it sets the section weights
The same structure serves very different pages depending on how aware the reader arrives
(seo-copy does the same move with search intent). Ask or infer, then say which stage
you're writing for:
- Problem-aware (cold traffic; they feel the pain but don't know solutions exist): the Problem/agitation section carries the page — spend words there, educate in "How it works," and don't rush the ask.
- Solution-aware (they're comparing approaches or vendors): lead with the differentiator and proof; compress the problem section to a nod — they already know.
- Product-aware (they know you; they arrived from a review, a demo, an ad): the page is about removing the last objections — weight FAQ, trust line, and friction-reducers, and put the ask high.
Mis-matching stage and structure is the classic silent failure: a definition-heavy page for a buyer who is ready, or a hard ask to a reader who doesn't yet feel the stake.
3. Decide the language, and route German early
If the page is German, read references/landing-de-notes.md before drafting — German
execution is native, not translated: native CTAs, du/Sie consistency, typography, no
Floskeln, Verbalstil. Do not write English copy and translate it; write German from the
meaning. For a bilingual page, build each language from its own craft.
4. Never fabricate proof
If a number, customer, or certification isn't real, write a clearly marked placeholder —
[X brands protected], [response time] — so the user drops in the real value. Invented
proof is the one unforgivable error: it ships as fact and destroys the trust the page exists
to build.
The section structure (overview)
Full detail — the job of each section, what good looks like, and bilingual notes — is in
references/page-structure.md. Read it before you plan a page. The default order,
adapt to the offer:
- Hero — benefit headline + subhead + primary CTA. The one promise, above the fold.
- Social proof strip — logos or a number, right under the hero, to buy a second look.
- Problem / agitation — name the reader's stake in their words, so the solution lands.
- Solution / how it works — usually three steps; turn the mechanism into a path.
- Features → benefits — each block leads with the outcome, names the feature, adds proof.
- Proof — case study, metrics, testimonials; the believability engine of the page.
- Objection handling / FAQ — answer the real reasons a ready buyer still hesitates.
- Trust line — checkable security/compliance facts, near the final ask.
- Final CTA — restate the one action with a friction-reducer.
Not every page needs all nine. A short page may fuse problem+solution and drop the FAQ. But the spine — hero → proof → solution → proof → ask — is near-universal. Cut sections; don't reorder into a structure that buries the point.
Process
- Confirm the one thing: the single primary action, and the one outcome the page sells.
- Plan the skeleton first: list the sections you'll use, each with its one idea, in one line. Show this to the user before writing full copy — it's cheap to fix here, expensive later. (See Example A.)
- Write section by section, front-loaded, each making one point, every CTA pointing at the one action.
- Apply the competitor copy-paste test to every line: could a competitor paste this sentence onto their site without it becoming false? If yes, it says nothing specific — replace it with a number, a named capability, a mechanism, or a proof point.
- Check the seams: consistent person/register throughout (in German, no du/Sie mixing), one primary CTA, proof real or placeheld.
- Run the cohesion pass (below): read the finished page top to bottom as one piece before delivering.
The cohesion pass — read it as one page
Section-by-section writing produces a page-level tell that no sentence-level check catches: every block shaped the same way. Read the assembled page top to bottom, aloud in your head, and fix:
- Repeated openings. Not every section (and not every feature block) may start with the same subject — "We … We … We …" or, in German, „Wir … Wir … Wir …“. Alternate the camera: the reader's moment, the mechanism, the outcome.
- Repeated shapes. If three consecutive headings follow the same pattern ("Verb your noun"), or every block is claim-then-proof in the same cadence, vary one.
- Word echoes. A key verb or noun ringing in every section ("protect … protect … protection") reads as a tic. Keep the brief's glossary terms; vary everything else.
- The heading skim. Read only the section headings: a scanning reader must still get the whole pitch, with no heading repeating another's job.
This pass changes little and catches what nothing else does — the assembled page is the unit the reader experiences, so it's the unit you check.
Worked examples
Example A — Input → Output: plan the skeleton from a sparse brief
Input (from the user):
"Landing page for our brand-protection tool. We find fake shops and phishing sites abusing a brand and get them taken down. Audience: security teams. We want demos booked. Proof: 1,200 brands, takedown confirmed in under 48h, GDPR-compliant, data in Germany."
Output (skeleton, shown before any full copy):
Primary action: Book a demo (every CTA points here).
- Hero — "Fake shops abusing your brand, taken down — not just reported." CTA: Book a demo.
- Social proof — "Protecting 1,200+ brands" + logos.
- Problem — finding out late, from an angry customer or the press; takedowns that drag.
- How it works — Detect → Report → Confirm (3 steps).
- Features → benefits — monitoring = know within the hour; takedown service = confirmed removal, not just an alert; weekly report = proof you acted.
- Proof — case metric: "[Customer] cut active impersonations by [X]% in [N] weeks."
- FAQ — "How fast is a takedown?" "What if it's a real reseller?" "Do you cover social?"
- Trust line — GDPR-compliant · data hosted in Germany · response within 24h.
- Final CTA — "See your brand's exposure in a 20-minute demo." Book a demo.
Confirm this shape and I'll write each section.
Why: the structure is settled and cheap to change before a word of body copy exists; one action repeats; every proof slot is mapped to a real, brief-sourced fact.
Example B — Input → Output: write one section (features → benefits)
Input:
"Write the features section. Product: staff scheduling for restaurants. Features: drag-and-drop shift planner; live labor-cost preview; absence requests in the app; weekly hours report. Audience: restaurant owners. Tone: warm, plain, no hype. Primary action: start a free trial."
Output:
Next week's schedule, done over one coffee. Drag shifts into place and publish — your team sees the plan on their phones the moment you hit save.
See what the week costs before you commit to it. The planner shows labor cost per day as you build, so the surprise is gone before it ever reaches payroll.
Absence requests stop living in your group chat. Staff ask for time off in the app; you approve with a tap, and the planner blocks those days automatically.
Every Monday: one short report — hours worked, overtime, open shifts.
CTA: Start your free trial · No credit card. Set up in an afternoon.
Why: the three headline blocks lead with the owner's moment of relief, then name the mechanism — while the weekly report is deliberately underweighted to a quiet fact line, because not every feature earns a bolded promise; the cost preview is proven by when it appears (before publishing), not praised as "smart"; no two blocks open with the same subject (the cohesion pass at work); the CTA points at the page's one action.
Hard rules vs. contested style
Hard rules (genuine errors — fix every time):
- Never fabricate proof. Bracketed placeholder instead.
- One primary CTA per view; every CTA points at the same action.
- Number & currency mechanics stay consistent across the page (group thousands; keep symbol and amount together). German specifics — currency after the amount, "20 %", no "in 2024", no Deppenleerzeichen, German quotation marks — are in the German notes; they are real errors, not preferences.
- Don't contradict the brief (voice, banned words, glossary, formality, du/Sie).
Contested style (smells — downgrade, never blanket-forbid):
These are tells of generic AI prose, not errors. Default to cutting them, but keep one if
it is genuinely the most accurate word. (Source: Wikipedia "Signs of AI writing"; see
humanizer-en / humanizer-de.)
- English AI-vocabulary: delve, leverage, robust, seamless, navigate the landscape, world-class, cutting-edge, best-in-class; em-dash overuse; rule of three on autopilot; "It's not just X — it's Y"; "studies show".
- German Floskeln & calques (handled in the German notes): innovativ, ganzheitlich, maßgeschneidert, "führender Partner"; "Sinn machen" (Duden marks it colloquial, but Eisenberg defends it — prefer "ergibt Sinn", don't call it wrong); "am Ende des Tages", "nicht wirklich".
Before you deliver
- Skeleton confirmed; each section makes exactly one point.
- Awareness stage named; section weights match it (problem-aware ≠ product-aware).
- One primary action; every CTA points at it; a friction-reducer sits by the final CTA.
- Every section front-loaded — the point is in the first line.
- Every claim has real proof or a marked placeholder; nothing fabricated.
- Every line passes the competitor copy-paste test.
- Voice matches the brief's adjectives and anti-adjectives — and mirrors the voice samples' rhythm where the brief has them.
- Cohesion pass done: varied section openings and shapes, no word echoes, and the heading skim alone still delivers the pitch.
- German page: read
references/landing-de-notes.md; du/Sie consistent, native CTAs, typography correct, no Floskeln, Verbalstil over Nominalstil.
References
references/page-structure.md— read before planning. The section-by-section structure: each section's job, the order, what good looks like, bilingual. Self-contained.references/landing-de-notes.md— read before writing a German page. Native German execution: CTAs, du/Sie, typography, Floskeln, Verbalstil. Self-contained, in German.- Optional enrichment (only if the full plugin is present; this skill works without them):
shared/references/conversion-principles.md,examples/before-after-en.md/examples/before-after-de.md, and theheadlines-and-ctas,copywriting-en,texten-de,typografie-de,du-sie-check,humanizer-en/humanizer-deskills.
What ships with it: 2 files
27.1 KB alongside SKILL.md
references/
- landing-de-notes.md12.6 KB
- page-structure.md14.5 KB
Gives 0 of the 12 instructions most design frontend skills give in ~3.3k tokens
Counted across 1,169 of the 1,878 authors here whose files we hold, read 2026-08-07
- Use CSS variables for color consistencyin 72 of 1169, across 23 files
- Commit to one bold aesthetic direction before codingin 72 of 1169, across 27 files
- Match implementation complexity to the aesthetic visionin 70 of 1169, across 20 files
- Add atmospheric background effects and texturesin 57 of 1169, across 9 files
- Use unexpected spatial compositions and layoutsin 56 of 1169, across 8 files
- Implement real working codein 55 of 1169, across 7 files
- Vary themes and aesthetics across different designsin 48 of 1169, across 7 files
- Launch chromium in headless modein 47 of 1169, across 4 files
- Close the browser when donein 47 of 1169, across 4 files
- Run provided scripts with help flag firstin 47 of 1169, across 4 files
- Wait for network idle statein 47 of 1169, across 4 files
- Use descriptive selectors for elementsin 47 of 1169, across 4 files
Said here and by no other author read
- ask for the primary action and outcome
- name the reader awareness stage
- plan the section skeleton before writing
- front-load every section
- make exactly one point per section
- point every call to action at one action
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.