millwright-labs /
minto-pyramid-skill
An Agent Skill that restructures your writing so the answer comes first
- 87 stars
- 9 stars since 3 Oct
- Updated 2 months ago
- Beginner friendly
- Agent skills & plugins
Models tend to write in narrative order: background, findings, then the point. That buries the one line a busy reader needs. This skill flips the order to the one consultants are taught: the answer, then the reasons, then the evidence under each reason.
It is one SKILL.md file with no dependencies. It defines five operations: restructure, buried-lede test, reason audit, so-what pass, and email version. It applies Minto's grouping rules to the middle layer and tells the model not to invent facts while tightening. The README notes Minto's rules do not require exactly three reasons, so the skill does not force that number.
It is deliberately restrained. When it loads on a draft you shared, it says what it noticed and asks before rewriting. It leaves incident timelines, runbooks, tutorials and thank-you notes alone. The README says this holds on Opus, Sonnet, GPT-5 and Gemini but not on Haiku, so use a stronger model or check the output.
What it does
Asks before rewriting
On a shared draft it gives a one-line observation and waits for a yes. It writes straight away only if you asked for the skill or asked it to write the document.
Five named operations
Restructure, buried-lede test, reason audit, so-what pass and email version. Ask for one by name or let it start with a restructure.
Grouping rules for the reasons
Each idea summarizes the ideas below it, and grouped ideas are the same kind and in logical order. SCQA is included too.
Declines the wrong documents
For timelines, runbooks and similar, it says so in a line and leaves them alone.
Documented limits and testing
The README lists a fact-check of its Minto claims, a with/without-skill test, and a known failure on Haiku.
What you could build
- Restructure a long recommendation email so the decision is in the first line.
- Run a buried-lede test on a draft memo to see where the point actually lands.
- Audit the reasons in a proposal to find overlapping or filler ones.
- Ask Claude to write a decision document for executives in answer-first form.
- Turn a rambling status update into a short email with the ask up front.
Get started
Clone the repo into your Claude skills folder (or just copy SKILL.md there) and restart your session. Then share a draft, or say: Use the minto-pyramid skill on this draft.
git clone https://github.com/millwright-labs/minto-pyramid-skill ~/.claude/skills/minto-pyramidCommands are copied from the project's README. Read it on GitHub for the full setup.
More in Agent skills & plugins
superpowers
A set of skills that makes coding agents plan, test and review before they write code
mattpocock/skills
Small, editable agent skills for planning, testing and reviewing code with AI
ECC
Agents, skills, hooks and memory that add an engineering workflow to coding agents
andrej-karpathy-skills
A CLAUDE.md file that tells Claude Code to ask first, keep code small, edit narrowly
anthropics/skills
Anthropic's example Agent Skills: folders of instructions Claude loads per task
ponytail
A skill that makes AI coding agents write the least code that works
gstack
Slash-command skills that give Claude Code roles like reviewer, QA and release engineer
addyosmani/agent-skills
A pack of 25 Markdown skills that guide AI coding agents from spec to ship
taste-skill
Agent skills that steer AI coding tools toward better layout, type and motion
archify
Agent skill that turns a description or a codebase into an interactive HTML diagram