https://github.com/escherize/pyramid-principle-skill
https://github.com/escherize/pyramid-principle-skill
Last synced: 9 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/escherize/pyramid-principle-skill
- Owner: escherize
- Created: 2026-06-17T22:28:14.000Z (18 days ago)
- Default Branch: master
- Last Pushed: 2026-06-17T22:43:37.000Z (18 days ago)
- Last Synced: 2026-06-18T00:16:32.989Z (18 days ago)
- Language: JavaScript
- Size: 16.6 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# pyramid-principle-skill
A Claude skill that makes Claude write and review prose, docs, memos, reports,
and slides in the structure of Barbara Minto's **Pyramid Principle**:
answer-first (BLUF), SCQA introduction, MECE grouping, and action-title headers.
It doubles as a self-enforcement loop that stops the model from building up to
the conclusion.
## Install
Use [`npx skills`](https://github.com/vercel-labs/skills) (the open agent-skills
CLI). It fetches `SKILL.md` and drops it into `.claude/skills/` (or the right
directory for whichever agent you use):
```bash
# into the current project
npx skills add escherize/pyramid-principle-skill
# into all projects (user-level)
npx skills add escherize/pyramid-principle-skill --global
```
### Without the skills CLI
Run this repo's own installer straight from GitHub:
```bash
npx github:escherize/pyramid-principle-skill # current project
npx github:escherize/pyramid-principle-skill --global # all projects
```
Or clone and run it directly:
```bash
git clone https://github.com/escherize/pyramid-principle-skill
node pyramid-principle-skill/bin/install.js --global
```
## Use
Once installed, ask Claude things like "structure this memo with the pyramid
principle", "make this more executive / lead with the conclusion", "is this
BLUF?", or just mention Minto / SCQA / MECE. The skill activates automatically.
## What it enforces
- **Governing thought first** — the single main idea as the opening sentence.
- **SCQA intro** — Situation, Complication, Question, Answer.
- **MECE groups** — no overlap, no gaps.
- **Three grouping rules** — same kind, summarizing parent, logical order.
- **Inductive > deductive** ordering at the top level.
- **Action titles** on slides — one message per slide, title states the conclusion.
- **30-second / first-sentence test** before shipping.
Derived from Minto, *The Minto Pyramid Principle* (1985/1987).
## Provenance
The framework rests on Minto's book, corroborated by McKinsey materials,
Wikipedia (MECE), and practitioner write-ups (strategyu.co, think-cell.com,
thinkinsights.net, modelthinkers.com).
The LLM-enforcement rules are grounded in: Anthropic prompt-engineering docs
(positive framing, anti-preamble, structured outputs), the reasoning-first vs
answer-first ordering result (dylancastillo.co; OpenAI structured-output docs),
negation-unreliability research (NeQA inverse-scaling; "Pink Elephant"
arXiv:2503.22395), and prompt-ordering effects ("A Taxonomy of Prompt Defects",
arXiv:2509.14404). The reason-hidden-then-BLUF two-stage pattern is a synthesis
of two verified findings, not a directly measured pattern.
## License
MIT