本文へ移動
cccskills
無料GitHub で公開

copywriting

Writes and rewrites the words on marketing pages: homepage, landing page, pricing, feature, product and about pages, including headlines, button text, section copy, title tag and meta description. Starts from the reader's own language and the proof the business can show, edits copy where it lives in the repository, and flags every claim that needs evidence. Use when someone says "write copy for this page", "rewrite our homepage", "this page sounds generic", "headline ideas", "better CTA", "landing page copy for this ad" or "pricing page wording". Email copy belongs to email-sequence, popups to popup-cro, and line-editing a finished draft to copy-editing.

インストール方法を見る

含まれるファイル(2)

  • SKILL.md14.6 KB
  • _scores.json1.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Copywriting

Overview

Page copy has one job: let a particular visitor work out, in the few seconds they give it, what this is, whether it is for them, and what to do next. Usability research has found for decades that most visitors scan a page instead of reading it, and that concise, scannable, plainly worded text is measurably easier to use than promotional prose. So the work is mostly decisions made before any sentence is written: who is reading, what they already know, what can be proved.

The skill returns copy placed where it will live (the component, the Markdown file, the translation file), a short brief that explains the choices, alternatives for the headline and button, and a list of claims the user must confirm before publishing.

Instructions

1. Find the copy and the constraints

Locate the page and any strings it pulls in before asking questions:

grep -rnE "<h1|<h2|title:|description:" src/app src/pages content 2>/dev/null | head -30
ls locales messages i18n 2>/dev/null

Read what is there, including neighbouring pages, so the new text matches the product's existing names for things. Then ask for what is missing:

  • The page and its one action. What should a visitor do here: start a trial, book a call, buy, join a list?
  • The reader and the door they came through. A search query, an ad (get the ad text), a link in an email, a referral.
  • The offer in plain terms: what it does, what it costs, what is free, what happens after clicking.
  • Proof on hand: customer numbers, measured results, named quotes with permission, awards, certifications.
  • Limits: brand voice notes, words legal has ruled out, regulated claims (health, finance, security).

2. Collect the reader's words

Copy sounds generic when it is written from the product's vocabulary. Ask for, or read in the repository and shared folders, any of these: support tickets, sales-call notes, reviews of this product and of its competitors, cancellation reasons, forum threads where the audience complains about the problem.

Pull four things out of that material and quote them exactly:

  1. The situation that sends someone looking ("every Sunday night I rebuild the rota").
  2. The words they use for the problem, which are rarely the category name.
  3. The result they want, in their terms.
  4. What makes them hesitate.

If no such material exists, say so and label the draft as a hypothesis to check against the first five customer conversations.

3. Write the brief first

Seven lines, agreed with the user before drafting:

Reader      who, in what role, at what size of company or stage of life
Situation   what just happened that made them look
Wants       the outcome in their words
Offer       the product and the mechanism that produces that outcome
Proof       what can be shown, with its source
Hesitation  the one or two doubts that stop the click
Action      the single thing this page asks for

Compress it into one sentence that the whole page defends, shaped like a Jobs to Be Done job story (situation, reader, outcome) with the mechanism added: "When next week's rota is still unfinished on Sunday night, a restaurant manager wants it done and seen by every employee, and Rotaleaf does that by sending a drag-and-drop rota to each phone." A page that cannot be summarised this way is trying to do two jobs; split it.

4. Lead with what this visitor needs to hear first

The visitor arrived byThey already knowOpen with
Searching the category ("staff scheduling app")They need this kind of productThe category named plainly, then what sets this one apart
Searching the problem ("staff keep missing shifts")The pain, not the product typeThe problem in their words, then the outcome
Clicking an adOnly the ad's promiseThe same promise, in nearly the same words, as the headline
An email or a referralThe brand, maybe the productThe specific news or offer
Comparing shortlisted vendorsThe alternativesThe concrete differences and the proof

5. Build the page as answers, in the order questions arise

Visitor's questionSectionWhat goes in it
What is this and is it for me?Headline, subheadline, primary buttonOutcome or category, who it is for, the next step
Can I believe it?Proof, straight after the opening blockA number with its unit and date, logos with permission, one named quote
What do I actually get?Three to five benefit blocksHeading states the result; body states how the product delivers it
How hard is it?StepsWhat the first ten minutes look like, numbered
What does it cost?Price or pointer to pricingThe number, the unit, what is included, what is not
What if it goes wrong?Objections, as questionsThe real hesitations from the brief, answered without hedging
What now?Closing blockThe offer restated in one line, the same button

Page types change the emphasis, not the logic. A homepage serves several kinds of visitor, so its headline is the broadest true statement and its sections route people onward. A landing page has one audience and one action: remove navigation and every sentence that does not serve that action. A pricing page answers "which plan is mine" (name the customer each plan fits) and "what will I really pay". A feature page starts from the task the feature completes. An about page explains why the company's history makes it a safe choice and still ends with a next step.

6. Write the lines

  • Headline: a visitor who reads only this line should know what is offered or what they will be able to do. Test: could a company in another industry use the same sentence? If yes, it says nothing.
  • Subheadline: adds the mechanism and the audience. One or two sentences.
  • Headings: each must make sense read alone, top to bottom, as a summary of the page.
  • Numbers: give the figure, the unit and the comparison ("payroll export in 4 minutes, down from an afternoon"). Round numbers read as invented; use the measured one.
  • Words: the reader's nouns, active verbs, "you" more often than "we". Cut intensifiers and any adjective the reader would not say aloud.
  • Buttons: say what happens on click ("Start 30-day trial", "See pricing"), and put the reassurance beside it in small text ("No card needed").
  • Link text describes its destination; "click here" fails both scanners and screen-reader users.

7. Keep every claim provable

Advertising law in the US requires that claims be truthful and that the advertiser hold evidence before publishing them; that covers what a sentence implies as well as what it states. In practice:

  • Every number carries a source the user can produce. Mark anything unverified as [CONFIRM: what is needed] in the draft.
  • Testimonials are real, attributed and used with permission. An exceptional result may not be presented as what customers generally get.
  • "Free" comes with its conditions next to it. Comparisons with competitors must be accurate on the day of publishing.
  • Drop "best", "leading", "#1" and "guaranteed" unless there is a document behind them.
  • No invented scarcity or countdowns, and no button text that shames the person who declines.

8. Write the search snippet

Give each page a unique title element and meta description. Google states no length limit for either and truncates to fit the device, so put the distinguishing words first and the brand last; as a rule of thumb, about 60 characters for a title and 150 for a description usually survive. The description is one or two sentences a person would want to click, not a keyword list. Google may show different text, so the visible headline has to work alone.

9. Check the draft

# stock phrases that carry no information
grep -nEio "seamless(ly)?|cutting-edge|world-class|revolutionary|next-generation|best-in-class|robust|leverag[a-z]*|unlock[a-z]*|supercharg[a-z]*|game-chang[a-z]*|empower[a-z]*|innovative|solutions?" content/home.md

# sentences longer than 25 words
grep -vE "^(#|[[:space:]]*$)" content/home.md | grep -oE "[^.!?]+[.!?]" | awk 'NF > 25 { print NF " words:" $0 }'

Then read the headings alone, read the page aloud once, and confirm each line of the brief appears on the page.

10. Deliver

  1. The agreed brief.
  2. The copy, applied to the files (show the diff), or as a labelled document when no code exists: title tag, meta description, H1, subheadline, button, button note, then each section.
  3. Two alternative headlines and one alternative button, each with the angle it takes and when to prefer it.
  4. The [CONFIRM] list: every claim awaiting evidence.
  5. What to test first, if the page has the traffic for a test.

Examples

Example 1: Landing page for a search ad

Request: "Write the landing page for our ad group 'restaurant staff scheduling app'. Rotaleaf, $3 per active employee a month, 30-day trial with no card. Three customer reviews attached."

The agent quotes the reviews ("I was rebuilding the rota in Excel every Sunday night"; "swaps happen in the group chat and I find out when someone doesn't show") and agrees this brief:

Reader      Owner or manager of an independent restaurant, 8 to 40 staff, builds the rota personally
Situation   Sunday night, redoing next week's rota in a spreadsheet
Wants       A rota finished quickly that staff really see, and no surprise gaps
Offer       Rotaleaf: drag-and-drop rota sent to staff by text link, optional app, swaps the manager approves
Proof       212 restaurants (company figure, June 2026); quote from Inés Calloway, Brasa & Vine
Hesitation  "My staff won't install another app"; "I have no time to set this up"
Action      Start the 30-day trial
Title tag:    Restaurant staff scheduling app | Rotaleaf
Description:  Build next week's rota from last week's, send it to every phone and approve shift swaps yourself. $3 per active employee. 30-day trial, no card.

# Restaurant staff scheduling without the Sunday-night spreadsheet

Build next week's rota, send it to every phone, and approve shift swaps before they turn into no-shows.

[Start 30-day trial]   No card needed. $3 per active employee after that.

Used by 212 restaurants (June 2026).

## Swaps you approve, not swaps you discover
Staff request a swap in the app. You get one notification, tap yes or no, and the rota updates for everyone.

## Next week's rota from last week's
Copy any week, drag the changes, publish. [CONFIRM: median time to publish a rota, from product analytics]

## "My staff won't install another app"
They do not have to. Everyone gets the rota by text message with a link; the app is optional.

> "I get my Sunday evenings back." Inés Calloway, owner, Brasa & Vine [CONFIRM: written permission]

Alternatives offered: "Stop rebuilding the rota every Sunday night" (leads with the pain; prefer for the problem-worded ad group) and "The rota your staff actually read" (leads with the result for managers whose trouble is no-shows).

Example 2: Replacing a hero that says nothing

Request: "Our homepage hero feels empty. File is src/components/Hero.tsx. Driftwatch watches Postgres schemas."

The check from step 9 finds five stock phrases in two sentences. The user confirms the facts: comparison runs every 60 seconds, alerts go to Slack, one database is free and needs no card. The agent edits the component and leaves the markup and class names alone:

@@ -3,7 +3,9 @@
     <section className="hero">
-      <h1>The next-generation platform to supercharge your database workflows</h1>
+      <h1>Know within a minute when production stops matching your migrations</h1>
       <p className="sub">
-        Driftwatch empowers engineering teams to seamlessly unlock visibility into their schema.
+        Driftwatch compares your live Postgres schema with the migrations in your repository
+        every 60 seconds and posts the exact difference to Slack.
       </p>
-      <a className="btn" href="/signup">Get started</a>
+      <a className="btn" href="/signup">Watch my first database</a>
+      <p className="note">Free for one database. No card needed.</p>
     </section>

It reports why: the old headline could sit on any software site; the new one names the event the reader fears and the time to detection, the subheadline gives the mechanism, and the button describes the first step. It adds a title tag ("Postgres schema drift monitoring | Driftwatch", 45 characters) and notes one item to confirm: that "within a minute" holds for the largest customer schema.

Guidelines

  • Do not draft before the action and the reader are known. Copy written for "everyone" has to be rewritten.
  • Never fabricate proof, and never soften an invented number into "up to". A gap in the evidence is reported, not papered over.
  • Changing words in code means changing only words: keep element structure, translation keys and variables intact, and update every locale file or flag the ones left untranslated.
  • Clever headlines cost comprehension. Wordplay is acceptable only when the literal meaning is still obvious to someone skimming.
  • A headline test needs traffic. On a page with a few hundred visitors a month, choose by the brief and by showing two versions to five people from the audience.
  • Length follows the decision: an expensive or unfamiliar product needs more answered questions, a free tool needs fewer. Cut sections, not clarity.
  • The legal notes describe US advertising rules in outline; regulated sectors and other countries need review by someone qualified.
  • Use a different skill for email sequences (email-sequence), popups (popup-cro), polishing existing text line by line (copy-editing), and diagnosing why a page underperforms beyond its wording (page-cro).

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Scripts and configures production rendering in Autodesk 3ds Max with the V-Ray and Corona renderers: output size and files, render elements, denoising, light mix, batch and command-line rendering, and network rendering. Use when a user asks to set up a production render, render several cameras in one batch, render from the command line or on a render farm, add render passes for compositing, or cut render time for archviz and product shots.

日本語の概要は準備中です。原文の説明を表示しています。

TerminalSkills/skills1652026年10月4日 更新

Covers scripting Autodesk 3ds Max, the 3D modeling and rendering application, with MAXScript and Python (pymxs): scene manipulation, object creation, material assignment, camera and light setup, batch operations, and file I/O. Use when tasks involve automating repetitive 3ds Max workflows, batch processing scenes, running scripts headless with 3dsmaxbatch, creating custom tools, or scripting scene setup for archviz, product visualization, or VFX.

日本語の概要は準備中です。原文の説明を表示しています。

TerminalSkills/skills1652026年10月4日 更新

3proxy

無料

3proxy is a small open-source proxy server that runs HTTP/HTTPS, SOCKS4/5, SNI and TCP/UDP port-mapping proxies from one config file. Use when a user asks to set up an HTTP or SOCKS5 proxy, add proxy users and passwords, write 3proxy access rules, chain or rotate upstream (parent) proxies, limit bandwidth, connections or monthly traffic per user, run 3proxy in Docker, or fix a 3proxy.cfg that will not start.

日本語の概要は準備中です。原文の説明を表示しています。

TerminalSkills/skills1652026年10月4日 更新

Builds Agent2Agent (A2A) servers and clients, the open protocol (originally from Google, now under the Linux Foundation) that lets AI agents from different frameworks call each other. Use when the user wants to create an A2A-compliant agent, build an Agent Card, implement task management, connect agents across frameworks, set up agent discovery, handle streaming responses, implement push notifications, or orchestrate multi-agent workflows. Trigger words: a2a, agent to agent, agent2agent, a2a protocol, a2a server, a2a client, agent card, agent interoperability, agent collaboration, multi-agent, agent discovery, a2a sdk, a2a task.

日本語の概要は準備中です。原文の説明を表示しています。

TerminalSkills/skills1652026年10月4日 更新

Plans a controlled experiment (A/B test) so its result can be trusted: writes the hypothesis, picks one primary metric and the guardrails, computes sample size and run time, specifies how visitors are assigned and when exposure is logged, and reads out the result with a confidence interval. Use when someone says "set up an A/B test", "split test this page", "how many visitors do I need", "how long should the experiment run", "is this result significant", "can I stop the test early", or wants to test a headline, price, layout or onboarding change against the current version.

日本語の概要は準備中です。原文の説明を表示しています。

TerminalSkills/skills1652026年10月4日 更新

ably

無料

Ably is a hosted realtime messaging service: clients publish and subscribe to named channels over WebSockets, see who is present, replay message history, and resume after a dropped connection. Use when a user asks to "add realtime updates", "push live notifications to the browser", "show who is online", "add a chat room with typing indicators", "publish from a serverless function", or "authenticate Ably clients without exposing the API key". Covers the ably 2.x JavaScript SDK (Realtime and REST), JWT token authentication, presence, history and rewind, batch publishing, and the @ably/chat 1.x SDK.

日本語の概要は準備中です。原文の説明を表示しています。

TerminalSkills/skills1652026年10月4日 更新

TerminalSkills のスキルをすべて見る

このスキルの問題を報告する