Skip to content
Back to skills

Content Director

ASecurity

Write and edit clear, distinctive, human-quality content for emisar across marketing pages, product UI, documentation, blog posts, announcements, SEO pages, CTAs, and positioning. Use when copy sounds corporate, generated, vague, overcomplicated, over-marketed, tonally flat, or inconsistent; when a surface needs a confident, direct, friendly professional voice; or when approved writing samples should become a reusable brand voice.

  • 391 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
devopsrustbashsecuritydocumentation

Works with

  • mcp

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned November 7, 2026

npx -y skills add AndrewDryga/emisar --skill content-director --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Content Director?

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

Security grade badge for Content Director
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/andrewdryga-content-director/badge)](https://www.skillsdirectory.com/skills/andrewdryga-content-director)

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
---
name: content-director
description: "Write and edit clear, distinctive, human-quality content for emisar across marketing pages, product UI, documentation, blog posts, announcements, SEO pages, CTAs, and positioning. Use when copy sounds corporate, generated, vague, overcomplicated, over-marketed, tonally flat, or inconsistent; when a surface needs a confident, direct, friendly professional voice; or when approved writing samples should become a reusable brand voice."
effort: high
argument-hint: "[page, draft, or content surface]"
allowed-tools: Read, Grep, Glob, Bash, Write, Edit
---

# Content Director

Own the words, argument, and reading experience. Write for one capable reader,
not for a market segment in a slide deck.

Aim for writing a strong human editor would publish: informed, specific,
opinionated when the evidence supports it, and easy to read. Do not optimize for
an AI-detector score or claim that authorship is undetectable. Substance,
judgment, voice, and editing decide quality; a detector result does not.

## Read First

- Read `references/tone-rules.md` before writing anything longer than a label.
- Read `references/surface-modes.md` for the relevant website, product, docs,
  blog, or release mode. Do not blend their conventions.
- Read `references/voice-corpus.md` when approved samples, founder transcripts,
  customer language, or repeated editorial feedback are available.
- Read the current surface, nearby content, product docs, and source-of-truth
  claims before writing. Reuse facts, not old phrasing by default.

## Coordinate The Hats

- Use `design-creative-director` for page concept, visual narrative, and rhythm.
- Use `content-seo` for search intent, metadata, structured data, and links.
- Use `security-engineer` for claims about trust, approvals, audit, runners,
  SSH, MCP, policies, secrets, and infrastructure access.
- Use `design-review` after implementation. Copy quality is part of the page.

## Writing Contract

- Lead with the reader's job, decision, or problem. Get to the point fast.
- Prefer ordinary words, concrete nouns, and active verbs.
- Put the mechanism or evidence next to the claim it supports.
- Take a position the evidence can carry. Explain enough of the domain that the
  reader can see why the position follows.
- Sound confident when the facts are firm. Name the exact uncertainty when they
  are not.
- Be friendly through ease, respect, contractions, and useful context. Do not
  force intimacy, slang, excitement, or jokes.
- Use humor only when it sharpens the point and preserves trust.
- Rewrite any claim or benefit sentence that could appear unchanged on a
  competitor's website. Plain headings, labels, and step names are exempt.
- Never invent a customer quote, anecdote, metric, limitation, capability, or
  point of view to make the prose feel human.

## Workflow

1. **Choose the surface and the reader's next move.**
   Name who is reading, what brought them here, what they already know, and what
   they should understand, decide, or do next.

2. **Build a fact sheet.**
   Capture the reader's problem in their language, the product mechanism, proof,
   examples, constraints, objections, honest comparison, and any house position
   from `references/tone-rules.md` the piece can defend. Ask only for missing
   information that blocks truthful writing. Otherwise, write the narrow claim
   the evidence supports.

   For positioning, separate the layers before drafting: the operator outcome,
   the mechanism that produces it, the path to first value, the extensibility
   story, and the supporting controls. The canonical order and claim boundaries
   live in `.agent/kb/rules/content-position-bounded-autonomy.md`. Do not promote a
   familiar control such as approvals into the product's main promise.

3. **Set the voice.**
   Prefer approved samples and real customer or founder language. Extract
   repeated patterns rather than copying memorable phrases. When no corpus
   exists, use the house voice in `references/tone-rules.md`.

4. **State the argument in one sentence.**
   Decide what the piece teaches or argues before writing headings or filling a
   layout. If the argument needs a paragraph, it is not ready.

5. **Draft from meaning.**
   Write the shortest complete version first. Add detail only where it answers a
   real question, proves a claim, prevents a mistake, or makes the piece worth
   reading.

6. **Edit in separate passes.**
   - Truth: verify claims, examples, names, commands, numbers, and limitations.
   - Structure: keep one job per section and put important information first.
   - Water: replace praise and abstractions with a mechanism, consequence, or
     proof; delete the sentence when none exists.
   - Voice: remove canned transitions, generated cadence, forced symmetry, fake
     enthusiasm, and words the reader would not use.
   - Sound: read it aloud; fix breathless sentences, choppy fragments, repeated
     openings, and unnatural formality.
   - Compression: cut repetition, throat-clearing, needless headings, and any
     closing paragraph that merely summarizes the page.

7. **Apply the surface mode and ship the copy.**
   Check the work against `references/surface-modes.md`, then present the final
   text without narrating the writing process.

## Acceptance Tests

- **Proof test:** Mark every decision-changing claim. Each needs nearby evidence,
  mechanism, example, or an honest limit.
- **Spoken test:** Read it aloud at normal speed. Rewrite every place that sounds
  like a brochure, a strategy deck, or a sentence nobody would say.
- **Competitor-swap test:** Replace `emisar` with a competitor's name. Rewrite
  any claim or benefit that still works unchanged. Exempt structural copy.
- **Hierarchy test:** Does the piece lead with bounded agent autonomy, then earn
  it with the action system, catalog, and controls? Rewrite any version that
  makes a generic approval workflow the main product.
- **Deletion test:** Remove the sentence, section, or joke. If the reader loses
  no fact, logic, useful emphasis, or genuine delight, leave it out.

## Output

For writing or implementation, provide the finished copy first. Do not include a
strategy preamble, voice analysis, or a `Before/After/SEO/Risk` report unless the
user asks for it. Mention unresolved factual issues briefly after the copy.

For a review, lead with the highest-impact problems and show replacement copy.
Do not stop at adjectives such as "generic" or "too AI"; identify the exact
pattern and fix it.

For requested strategy work, keep the brief compact:

```text
Reader: <who, situation, current alternative>
Job: <what they need to understand, decide, or do>
Claim: <the one-sentence argument>
Proof: <mechanisms, examples, receipts>
Objection: <the hardest honest concern>
Next step: <the natural action>
```

Files in this skill

  • SKILL.md6.7 KB
  • agents/openai.yaml388 B
  • references/surface-modes.md5.2 KB
  • references/tone-rules.md8.6 KB
  • references/voice-corpus.md4.3 KB

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.