A Claude Code skill that generates Amazon-style decision documents — narrative 6-pagers and PRFAQs (Press Release + FAQs) — with an embedded prose audit grounded in Strunk & White and Anthropic's removability test.
The format Amazon has used since 2004, when Bezos banned PowerPoint and required every meeting to open with a written narrative.
Three independent traditions converge on a single principle: writing is the thinking instrument, not the documentation of it.
- Bezos (2004): "There is no way to write a six-page narratively structured memo and not have clear thinking."
- Strunk (1918): "Omit needless words. Vigorous writing is concise."
- Anthropic: "For every line, ask: If I removed this, would Claude make a mistake? If not, remove it."
Same rule, three scales — line, sentence, instruction. This skill applies all three.
/six-pager memo → 6-page narrative memo (decisions, strategy, status)
/six-pager prfaq → Press Release + External FAQ + Internal FAQ (launches)
Auto-detect: if input contains "launch" / "ship" / "announce" / "new product" → suggests prfaq. Otherwise → memo.
git clone https://github.com/ngmeyer/six-pager.git
cd six-pager
# Global install (available in all projects)
mkdir -p ~/.claude/skills/six-pager
cp SKILL.md ~/.claude/skills/six-pager/SKILL.md
# Or per-project install
mkdir -p .claude/skills/six-pager
cp SKILL.md .claude/skills/six-pager/SKILL.mdUse it in Claude Code:
/six-pager Should Voltron use half-Kelly sizing?
/six-pager memo docs/q3-strategy-draft.md
/six-pager prfaq SignUpSpark Sparky AI public launch
/six-pager --silent-read docs/voltron-decision.md # generates margin questions
/six-pager --strunk-only docs/draft.md # audit only, no new structure
Trigger phrases: six pager, 6-pager, amazon memo, narrative memo, PRFAQ, press release, work backwards, decision memo, strategy doc, launch document.
| Section | Length | Purpose |
|---|---|---|
| Introduction | ~0.5 pg | Document purpose, audience, ask |
| Goals | ~0.5 pg | Specific, measurable outcomes |
| Tenets | ~0.5 pg | Principles this proposal optimizes for (no qualifiers) |
| State of the Business | ~1.5 pg | Current data, named systems, concrete numbers |
| Lessons Learned | ~1 pg | What past attempts taught (specific events) |
| Strategic Priorities | ~2 pg | Recommended action, milestones, owners |
| Appendix | unlimited | Visuals, alt analyses, supporting data |
Hard cap: sections 1–6 fit in 6 pages of 11pt body text.
1. Press Release (~1 pg)
- Customer-facing, benefit-led, dated for launch day
- Headline + sub-headline + lede + problem + solution + quotes + CTA
2. External FAQ (~1 pg)
- What journalists / customers would ask
- Differentiation, pricing, availability, what it doesn't do
3. Internal FAQ (~1-2 pg)
- What leadership would ask
- Why now? Success metrics? Failure mode? What would kill this?
Work backwards — write the press release first. The PRFAQ test: if you can't write a credible press release, the product doesn't exist yet.
Every document gets a Strunk + Anthropic audit. Findings categorized:
| Check | Rule | What it catches |
|---|---|---|
| Passive voice | Strunk R14 | "It was decided" → "We decided" |
| Vague language | Strunk R16 | "improvements" without a number |
| Needless words | Strunk R17 + Anthropic | "in order to" → "to" |
| Qualifiers | E.B. White | "rather," "very," "somewhat" — flag for deletion |
| Parallel construction | Strunk R19 | Inconsistent grammatical form in lists |
| Sentence-length monotony | Strunk R18 | 4+ same-length sentences in a row |
| Hedged Tenets | Bezos canon | "We try to" / "We strive to" — cut or rewrite |
| Removability | Anthropic | Paragraph that doesn't change reader's decision → cut |
| Flag | Effect |
|---|---|
--silent-read |
After generation, produce an annotated version with 2-4 margin questions per section — simulates the 20-30 minute Amazon meeting silent-read |
--strunk-only |
Skip structural generation; treat input as existing draft and run only the prose audit |
| Skill | Use For |
|---|---|
/six-pager memo |
Decisions, strategy, status, "should we?" with stakes |
/six-pager prfaq |
Product / feature / launch ideas where customer outcome is the question |
/article-draft |
PithyByte article in Neal's voice |
/human-writing |
Voice / anti-AI-tells / banned-words audit (any prose) |
/council-review |
Open question with multiple valid answers (deliberation, not generation) |
/adversarial-review |
Stress-test a finished six-pager (or any artifact) |
Natural pairing: generate with /six-pager, stress-test with /adversarial-review, deliberate the recommendation with /council-review. Full Amazon-style decision pipeline.
A six-pager passes if every check is true:
- Page count ≤ 6 (memo) or correct sub-document length (prfaq)
- Every Goal has a number or verifiable outcome
- Every Tenet is a declarative commitment with no qualifiers
- State of Business cites ≥5 specific numbers / named systems / dated facts per page
- Strategic Priorities name at least one owner per priority
- Strunk audit: zero critical findings
- "What would cause us to kill this?" (PRFAQ Internal FAQ) has a real, specific answer — not "if performance suffers"
- Exceed 6 pages. The constraint IS the value.
- Generate fake numbers. If a Goal needs a metric and the user hasn't supplied one, the skill asks.
- Hedge Tenets. A Tenet with "try to" is not a Tenet.
- Skip the audit. Prose discipline is not optional.
- Auto-save. Always confirms where to write.
- Generate prose articles. Use
/article-draftor/human-writinginstead.
- Bryar & Carr — Working Backwards: Insights, Stories, and Secrets from Inside Amazon (St. Martin's Press, 2021)
- Strunk & White — The Elements of Style, 4th edition (Pearson Allyn & Bacon, 2000)
- Tufte — The Cognitive Style of PowerPoint (2003)
- Anthropic — Claude Code Best Practices — the removability test
- Slab — How Bezos Turned Narrative Into Amazon's Competitive Advantage
- Format: Amazon's 6-pager and PRFAQ practice, established 2004
- Prose discipline: Strunk & White (1918, 1959, 2000)
- Removability test: Anthropic Claude Code documentation
- Skill by: Neal Meyer
MIT