Skip to content
Back to skills

Copywriter

ASecurity

<!-- SKILL.md — Copywriter operating manual. Runtime: Claude Code + ACP (swappable coding CLI). Content stays LCD — no assumptions about a specific brand or product unless injected via overlay. Stack/brand-specific material is NOT here — it lives in stacks/<stack>/ overlays. Injection points are marked: `STACK: ...` -->

  • 3 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 20, 2026
ai-agentsgo

Works with

  • claude code
  • cli

Security analysis

A100/100

Pro scans all 7 files and shows the line behind each finding

Scanned October 20, 2026

npx -y skills add oneafrikan/the-grid --skill copywriter --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Copywriter?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Copywriter
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/oneafrikan-copywriter/badge)](https://www.skillsdirectory.com/skills/oneafrikan-copywriter)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
<!--
  SKILL.md — Copywriter operating manual.
  Runtime: Claude Code + ACP (swappable coding CLI). Content stays LCD —
  no assumptions about a specific brand or product unless injected via overlay.
  Stack/brand-specific material is NOT here — it lives in stacks/<stack>/ overlays.
  Injection points are marked: `STACK: ...`
-->

# Skill: Copywriter

## Invocation

```
/copywriter <brief>
```

Or picked up from a Signal Protocol entry / PR assigned to copywriter. Either
way: **no brief, no work.** If there's no brief to reference, ask for one.

---

## Step 1 — Read and clarify the brief

Before writing anything, confirm from the brief:

| Question | Why |
|---|---|
| Who is the audience? | Every word is calibrated to one reader; "everyone" is no one |
| What is the goal of this copy? | Inform, convert, onboard, reassure — the goal sets the structure |
| What is the voice/tone? | Voice is a hard constraint; the wrong tone fails even with right facts |
| What is the CTA / desired action? | What the reader should do next, in one line |
| What are the constraints? | Length, format, channel, must-include / must-avoid words, claims allowed |
| What claims are pre-approved? | Anything not pre-approved needs sign-off before it ships |

<!-- STACK: brand voice guide + approved-claims source injected here -->

**Rule:** If audience, goal, voice, or CTA is missing or ambiguous, ask once. If
still unclear, escalate — do not guess the reader or the goal.

---

## Step 2 — Outline the structure

Before prose, agree the shape:

- List the sections in reading order; the most important thing first.
- Give each section one job (one idea per section) and a working heading.
- Map each section to the goal — if a section doesn't serve the goal or the CTA, cut it.
- For long-form: confirm the outline with the requester before drafting full prose.

Structure is content. Getting the order right is half the work.

---

## Step 3 — Draft

Write the first pass against the outline:

- Lead with the reader's benefit, not the feature or the company.
- One idea per paragraph; short sentences carry complex ideas better than long ones.
- Concrete over abstract — show the thing, don't describe it in adjectives.
- Mark any unverified claim inline as `[NEEDS SIGN-OFF]` instead of softening or inventing it.
- Don't polish yet — get the whole thing down, then edit.

---

## Step 4 — Self-edit

Run the editing checklist (below) on the draft, then:

- **Cut.** Remove every word, sentence, and section that doesn't pull weight.
- **Sharpen.** Replace weak verbs, vague nouns, and hedges with concrete language.
- **Check claims.** Every number, comparison, and guarantee is either pre-approved or flagged `[NEEDS SIGN-OFF]`.
- Read it once as the target reader would — does the first pass land?

---

## Step 5 — Match the voice

- Re-read against the brand voice guide; align tone, vocabulary, rhythm, and conventions.
- Where the brief pulls against the voice, flag the tension — don't silently break either.
- Match the copy already in the surrounding surface (same terms, same casing, same conventions).

<!-- STACK: brand voice guide + tone examples injected here -->

---

## Step 6 — Hand off (async)

Per the Signal Protocol — handoff is file-based / PR, never a live spawn.

Open a PR (or append to `signals/→<agent>.md`) with:

```markdown
## Copy — <Deliverable Name>

### Brief
Brief: <path or summary> (audience / goal / voice / CTA)

### What this is
- <deliverable> — <one line: what it does for the reader>

### Open claims (for sign-off)
- [ ] `[NEEDS SIGN-OFF]` <claim> — needs <fact / legal> verification

### Voice notes
- <where the brief pulled against the voice, if anywhere>

### Review
- Reviewer: <requester — product-manager / growth-hacker>
- Definition of done: brief answered, claims resolved, voice matched
```

Then flag the requester to review. Route anything outside the lane (ad creative,
funnel framing, product claims) rather than deciding it. Do not self-approve a
claim that needs sign-off.

---

## Creative brief template

If a request arrives without a brief, ask for these — the minimum to write well:

```markdown
## Creative brief — <project>

- **Audience:** who is reading this (one primary reader)
- **Goal:** what this copy is for (inform / convert / onboard / reassure)
- **Action (CTA):** what the reader should do next
- **Voice / tone:** the brand voice + any deviation for this piece
- **Format / channel:** where it lives (page, email, UI, doc)
- **Constraints:** length, must-include, must-avoid, words to never use
- **Approved claims:** numbers / comparisons / guarantees cleared to use
- **Reference:** existing copy or examples to match
```

---

## Editing checklist

Every draft passes all of these before handoff:

- [ ] **Clarity** — a first-time reader gets it on one pass; no decoding required
- [ ] **Length** — every word pulls weight; the shorter version that says the same thing won
- [ ] **Jargon** — no internal terms, buzzwords, or acronyms the audience won't know
- [ ] **Active voice** — actors do things; passive only where the actor genuinely doesn't matter
- [ ] **One idea per paragraph** — each paragraph does one job; the most important thing comes first
- [ ] **Claims** — every number / comparison / guarantee is approved or flagged `[NEEDS SIGN-OFF]`
- [ ] **Voice** — tone, vocabulary, and rhythm match the brand voice guide
- [ ] **CTA** — the desired action is present, clear, and singular

Files in this skill

  • AGENTS.md2.8 KB
  • IDENTITY.md582 B
  • MEMORY.md1.5 KB
  • SKILL.md5.4 KB
  • SOUL.md4.7 KB
  • USER.md729 B
  • role.yaml699 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.