Content Writer

Before you write anything

Read these first. They are the source of truth and this skill does not invent around them.

1. brand-kit/brand.md: name, palette, fonts, CTAs
2. brand-kit/voice.md: house style and banned constructions
3. brand-kit/positioning.md: the frame, pillars, traps, stats, vocabulary
4. brand-kit/capabilities.md: read before making any product claim

Three shared references carry the structure this skill applies. Read the ones the deliverable needs:

5. docs/writing-templates.md: the asset skeletons and every length target
6. docs/aeo-style-guide.md: how to structure anything meant to be quoted by an answer engine
7. docs/messaging-patterns.md: the eight structural patterns, one per asset

If brand-kit/ is missing or still has its template placeholders, stop and tell the user to run brand-kit-setup. Do not proceed with invented positioning.

1. How this skill works

1. Read the canon. brand-kit/positioning.md is the source of truth for the frame, the qualities, named traps, quotes, taglines, stats, and vocabulary. Every claim in the deliverable traces back to it.
2. Read the guardrail. brand-kit/capabilities.md defines what the product does and does not do. Check it before writing any capability claim. An inaccurate claim costs more credibility than a weak sentence.
3. Identify the deliverable. Blog post, landing page, email, case study, whitepaper, executive brief, feature page, pillar article, or supporting article.
4. Pick the lead quality. Most assets anchor one of the qualities in The Pillars, even when all of them show up. This is an internal decision. It shapes what you emphasize; it never appears in the copy.
5. Apply the asset template named in Asset Templates below. The shape and the length come from docs/writing-templates.md, and for anything written to be retrieved and quoted, docs/aeo-style-guide.md governs on top of it.
6. Pick the pattern. One structural pattern from docs/messaging-patterns.md carries the argument. Choose it before drafting, not after.
7. Apply voice. Follow Compact Voice Rules. If house-style is installed, run it as the final pass.
8. Deliver. End with a short note naming the lead quality and the target audience, kept outside the asset itself so it never ships to a reader.

If the user only asks a positioning question, "how do we differentiate against platforms that do X?", answer straight from brand-kit/positioning.md. Do not produce a full asset nobody asked for.

2. Naming and usage rules

Absolute. Violations are revision triggers, not style notes.

• Never name the framework. The qualities in The Pillars are internal scaffolding. "Our pillars," "our framework," and the qualities recited as a named set do not appear in customer-facing copy. The reader absorbs the qualities through the writing. This is also penalty 23 in the house-style humanizer rubric, so exposing it costs points twice.
• Follow Naming and Usage Rules in brand-kit/positioning.md for capitalization, always-say and never-say lists, and product-name formatting.
• Use taglines verbatim or not at all. They are in the same section.
• Use the product pronoun set in Company in brand-kit/brand.md. Never mix pronouns inside one asset.
• Check Superseded Names in brand-kit/voice.md before writing any product name. A rename in the kit propagates everywhere.
• Quote verbatim with attribution from External Validation Quotes or Internal Quotes in brand-kit/positioning.md. Never paraphrase a quote, never shorten one to make it fit, and never use an internal quote that is not marked cleared for external use.
• Cite every stat from Headline Stats. A number without a source-column entry does not ship. If Verified is flagged, say so in your delivery note and let the user decide.

3. Sell the idea, not the product

The core writing principle. The best content sells the idea that leads the reader to you, rather than pushing you at the reader. Position the product as the natural consequence of sound thinking.

In practice:

• Spend the first third of the asset on the reader's problem, with no product in it. If the opening still works with the company deleted, that is the correct opening for a reader who does not know you yet.
• Introduce the product where it answers a question the reader is now asking. Not before.
• Argue for the approach, not the vendor. A reader convinced of the approach will shortlist you. A reader pitched too early will not finish.
• Never resolve the argument with the contrast formula. "Most companies do X. Winners do Y." and "It's not about X, it's about Y." are banned outright in brand-kit/voice.md. State the failure and its consequence directly instead.

4. Asset Templates

The skeletons and the length targets live in docs/writing-templates.md. The answer-engine rules live in docs/aeo-style-guide.md. This skill restates neither. What follows is which template each deliverable uses, which guide governs its structure, and what to pull from the kit for it.

Every length comes from Length targets in docs/writing-templates.md. Never set a word count from memory.

4.1 The answer-engine rules, applied

Any row marked above as governed by docs/aeo-style-guide.md carries the elements below on top of its template beats. The rules themselves are in that file, under the headings named. Read them there and do not paraphrase them here.

1. Answer the title inside the opening. The 100-word rule.
2. Question-based H2s. Question-based headings. Phrase every major heading as a question a buyer would actually type, using the terms in Canonical Vocabulary in brand-kit/positioning.md.
3. An atomic answer under every question heading. The atomic answer rule. Write it first and the section around it second.
4. A Key Takeaways block after the intro, above the first H2. The Key Takeaways block.
5. An FAQ section at the end. The FAQ section. Source the questions from real sales objections and real search queries, never invent them.
6. The Publication checklist before delivery, including its internal linking rule, which matches the descriptive anchor rule in Compact Voice Rules below.

These do not apply to emails, case studies, executive briefs, or cold outreach. A person opens those. An engine does not retrieve them.

4.2 Blog post or article

Template: Article.

Pull from: the sections of brand-kit/positioning.md relevant to the topic. Always include at least one entry from Named Traps, at least one tagline or signature line from Naming and Usage Rules, and one quote from External Validation Quotes when the audience warrants it. Use Categories We Are Not for competitive contrast without naming competitors gratuitously. Proof comes from Headline Stats, and the close is the blog CTA from Calls to Action in brand-kit/brand.md.

4.3 Landing, feature, or solution page

Template: Landing page or feature page.

Pull from: the relevant quality's block in The Pillars, Named Traps, Categories We Are Not, Headline Stats, and the matching What We Do block in brand-kit/capabilities.md. Social proof is a cleared quote or the placeholder [Customer quote / case study reference]. The CTA is the landing-page primary from Calls to Action.

4.4 Email

Template: Email.

Pull from: the topic-relevant sections of brand-kit/positioning.md and the email CTA from Calls to Action. Subject line patterns are in docs/writing-templates.md as well.

4.5 Case study

Template: Case study.

Pull from: the customer data the user provides, plus the matching capability block in brand-kit/capabilities.md for an accurate description of what was deployed. Results are hard numbers: time saved, cost reduced, throughput increased, cycle time cut. The quote is verbatim or a clearly marked placeholder.

4.6 Executive brief and whitepaper

Templates: Executive brief for the one-screen version, Whitepaper for the long argument.

Pull from: The Manifesto, The Frame, Named Traps, Headline Stats, and the relevant blocks of The Pillars. The manifesto is what the executive summary and the recommendation argue from. A brief that only inventories capabilities gives the reader nothing to decide. Market context uses Why now from The Frame.

4.7 Pillar article

Template: the Pillar guide template in docs/aeo-style-guide.md, applied in full, including the recording-to-pillar method when the source is a transcript.

Pull from: brand-kit/positioning.md broadly, and The Manifesto first. The pillar piece is definitional: The Manifesto supplies the argument and The Frame is its spine. Use the canonical noun and its canonical definition from The Frame verbatim, treat the qualities as the standard any real solution must meet without naming them, and close on the sequence in Where to Start. Thought leadership that does not carry the manifesto's point of view is a summary of the category, which every competitor has already published.

4.8 Supporting article (cluster page)

Template: Article, with the answer-engine rules in 4.1 applied.

Pull from: the topic's coverage in brand-kit/positioning.md, plus explicit cross-links to the pillar article and to the pages listed in Link Map in brand-kit/capabilities.md.

5. Compact Voice Rules

The full rules live in brand-kit/voice.md, and house-style enforces them. The rules below are the minimum that must hold in any output from this skill even when it runs alone.

• No em-dashes. None. Not one.
• No appositive phrases that restate the subject.
• No passive voice.
• Action verbs early in the sentence.
• Paragraphs of one to four sentences.
• No banned vocabulary. The list is Banned Words and Phrases in brand-kit/voice.md.
• No contrast formula. "Most companies do X. Winners do Y." and "It's not about X, it's about Y." are both banned.
• No "in today's..." openers. No "imagine if." No "picture this."
• No concluding paragraph that summarizes what was just said. End on the CTA, a question, or the sharpest line in the piece.
• Numbers, not adjectives. Every number sourced from Headline Stats.
• One structural pattern from docs/messaging-patterns.md is visible in every customer-facing asset. Two stacked in a short asset is one too many.
• Vary section structure. If every section runs problem, solution, product, benefit, the piece reads as assembled. Open one section on a stat, one on a scene, one on a contradiction.
• Link inline with descriptive anchor text. Write "the platform completes third-party risk assessments in minutes," never "learn more: URL." Destinations come from Link Map in brand-kit/capabilities.md.

6. Workflow

Step 1, Gather inputs

Ask for anything you cannot infer:

1. Content type: blog post, landing page, email, case study, whitepaper, executive brief, feature page, pillar article, supporting article.
2. Topic or source: a URL to expand, a topic brief, an outline, or a specific capability.
3. Lead quality: which of the qualities in The Pillars anchors the piece. Internal only. It never appears in the copy.
4. Target audience: which role group from Buying Committee in brand-kit/icp.md, and which tier of The Fit Matrix.
5. CTA: which context from Calls to Action in brand-kit/brand.md.
6. Length: the default is the figure for this asset type in Length targets in docs/writing-templates.md. Ask only if the user wants to override it, and say what the default was.

If the user gives a URL, read it first and extract the core argument, the usable numbers, and the intended audience before writing a word.

Step 2, Check the canon and the guardrail

Confirm, before drafting:

• Which pattern from docs/messaging-patterns.md carries the argument. One per asset.
• Which entries in Named Traps fit this topic.
• Which quote from External Validation Quotes reinforces the angle, if any.
• Which tagline from Naming and Usage Rules belongs in this asset.
• Which numbers from Headline Stats apply, and whether each is verified.
• Which capabilities the asset will claim, and whether each one is accurate per What We Do and permitted by What We Do Not Do in brand-kit/capabilities.md.
• Whether the topic appears in Write About, Never Claim. If it does, explain the concept and stop there. Do not connect it to the product.

Step 3, Apply the template

Use the row for this deliverable in Asset Templates, read the named template in docs/writing-templates.md, and pull exactly what the Pull from line names. Where the row names docs/aeo-style-guide.md, apply the elements in The answer-engine rules, applied. Do not substitute memory for either file.

Step 4, Write

Apply Compact Voice Rules. Check every paragraph against this list:

• Is the subject acting?
• Did I avoid appositives?
• Are there zero em-dashes?
• Is the verb strong and specific?
• Does this paragraph earn its place, or does it restate the one above?
• Does this section open differently from the previous one?
• Am I selling the idea and letting the product be the conclusion?
• Is every capability claim accurate per brand-kit/capabilities.md?
• Is every number in Headline Stats with a source?
• Have I named the framework, the pillars, or the qualities as a set anywhere? If yes, cut it.

Use the exact terms from Canonical Vocabulary in brand-kit/positioning.md. Never the alternatives listed in its Never say instead column.

Step 5, Pre-delivery scan

Before delivering, scan the entire output for:

1. The , character.
2. The -- sequence used as an em-dash.
3. Any term from Banned Words and Phrases.
4. Any number that is not in Headline Stats.
5. Any quality name, "pillars," or "framework" in the body copy.
6. Any claim that contradicts What We Do Not Do.
7. Any bare "learn more: URL" link, and any anchor that is a bare keyword rather than a descriptive phrase inside a sentence.
8. On an answer-engine asset: a heading that is not a question, a section whose first paragraph does not stand alone, a missing Key Takeaways block, or a missing FAQ.

Replace every instance. There are no exceptions to the em-dash rule.

Step 6, Format and deliver

Clean markdown:

• H1 for the title, H2 for major sections, H3 sparingly.
• Sentence case in headings, not Title Case.
• Bold for key phrases, never whole sentences.
• Numbered lists for processes, bullets for parallel options, prose everywhere else.
• Zero em-dashes anywhere in the output.

Above the asset, in a delivery note the reader never sees, state the lead quality, the target audience, the traps and stats used, and any stat that is still flagged unverified.

If house-style is installed, run it as the final pass and report the score.

7. Notes

• Sourced numbers always outperform generic claims. If the source material has numbers, surface them early and cite them.
• The qualities are invisible scaffolding, not a checklist to recite.
• Do not overuse one tagline. A signature line that appears in every asset stops reading as a signature and starts reading as a tic.
• Always tie back to a business outcome: reduced cycle time, reduced cost, increased throughput, improved compliance, increased revenue.
• For search-driven educational content, some topics exist to earn traffic even though the product does not address them. Explain the concept and move on. Never write "we do not do this." The reader does not need the disclaimer, and Write About, Never Claim in brand-kit/capabilities.md already draws the line.
• When a topic needs a claim the kit does not support, stop and tell the user which brand-kit heading needs an entry. Do not write around the gap.

8. Cross-References

All optional. This skill works alone.

• brand-kit-setup: run it if the kit is missing or still has placeholders.
• house-style: if it is installed, run it as the final pass on any customer-facing asset, and use its quality panel before publishing.
• social-generator: if it is installed, use it for social posts instead of this skill.
• transcript-to-article: if it is installed, use it when the input is a raw transcript rather than a topic.
• competitive-white-space: if it is installed, use it to find the angle before writing a competitive piece.
• brand-visuals: if it is installed, use it for the featured image that goes with an article, and for a branded PDF of a whitepaper or executive brief.

Anything not in skills/ does not exist. Do not reference it.