Minto

Need a document someone can decide from?

Settle the argument before rewriting the prose.

Minto makes Codex and Claude Code identify the reader's question before they edit the wording. The finished draft puts the answer first and shows where the support is weak.

Install Minto See a before-and-after example
Answer Approve the pilot now.
ReasonGives us the missing data
ReasonWorks with our current process
ReasonKeeps the test small
Three comparable reasons support one answer.

Install Minto in two commands.

Add the marketplace and install the plugin. Start a new session after installation.

Codex

CLI & desktop app
  1. codex plugin marketplace add welltraum/minto
  2. codex plugin add minto@minto

Then invoke $minto or describe the writing problem directly.

Claude Code

CLI & desktop app
  1. /plugin marketplace add welltraum/minto
  2. /plugin install minto@minto

Then invoke /minto:minto or let Claude route a matching request.

The original update hides its recommendation in the last sentence.

This fictional example keeps the same facts, but puts the decision first and shows how each fact supports it.

Project update Topic first

We have been looking at the onboarding problem.

Research ran five interviews, design prepared two directions, and engineering estimates three weeks.

Activation remains below target.

RecommendationThe team would like to run a pilot next month.

The reader has to infer the point from the background.
Project update Answer first

RecommendationApprove a three-week onboarding pilot next month.

  • 01Activation remains below target.
  • 02Research completed five interviews.
  • 03Design prepared two directions.
The same facts now support one visible decision.

Use the mode that matches the job.

One source can produce four different outcomes. Choose the mode by what you need to know or make next.

intent

Decide what the document must achieve before drafting it.

Identify the reader, the decision, and an answer that the available evidence can support.

Reader question

Should we run an onboarding pilot next month?

One-sentence answer Approve a three-week pilot to test the two prepared directions. Evidence gap

Define the activation target before launch.

audit

See where the logic breaks without changing the source.

Minto scores the structure, names the defect, and prescribes a fix. The original wording stays intact.

4.0Structure score
  1. Finding The recommendation appears last.
  2. Finding Evidence and actions sit at the same level.
  3. Fix Lead with the pilot decision, then group the evidence.

write

Turn notes into a document that starts with the conclusion.

Use it for a memo or email, as well as one-pagers, updates, and deck storylines.

Approve a three-week onboarding pilot next month.

Activation is below target. Research completed five interviews, and design prepared two directions. Engineering estimates three weeks.

viz

Make the pyramid visible when prose hides the structure.

Render the pyramid and SCQ flow in Mermaid or build a standalone interactive page for sharing.

Approve the pilot
  • Need
  • Readiness
  • Scope
  1. Situation What the reader already accepts
  2. Complication What changed or now blocks
  3. Question What the reader must decide
  4. Answer What they should conclude or do
SCQ brings the reader to the question. The pyramid shows the answer and its support.

Start with the draft you actually have.

Minto can work from an unclear goal or a draft that still feels wrong. You do not need to clean it up first.

Release benchmark, v1.5.0

Quality rose by 1.13 points on average

Across Codex and Kimi, the skill arm scored 6.75/10 versus 5.63/10 for the control in the v1.5.0 release benchmark.

Mean quality score out of 10
Control
5.63
With Minto
6.75

Change by engine

Codex quality+1.25
Codex structure+1.38
Kimi quality+1.00
Kimi structure−0.38
View the benchmark table
Average benchmark results across eight fixtures
Engine Quality Δ Structure Δ
Codex +1.25 +1.38
Kimi +1.00 −0.38
Both engines +1.13 +0.50

32 evaluated cells, with one run for each engine, arm, and fixture combination. Kimi's structure score fell by 0.38 even though its overall quality score rose. The benchmark shows an effect, but it does not guarantee improvement in every case.

Read the full methodology and limitations

Use Minto on the draft you have.

Install the plugin, paste your text, and ask Minto to fix the structure before it edits the wording.

Choose your install