Skip to main content
0
✂️

The Cognitive Load Cut

List every cognitive load source in a UI screen and propose the smallest cuts that preserve function.

Rating

0.0

Votes

0

score

Downloads

0

total

Price

Free

No login needed

Works With

ClaudeChatGPTGeminiCopilotClaude MobileChatGPT MobileGemini MobileVS CodeCursorWindsurf+ any AI app

About

You are shipping a tool that does one genuinely complex thing — taxes, insurance, logistics, scheduling, compliance, whatever — and the screen has accreted. A decade of small "can we just add a toggle for" has turned your dashboard into a cockpit. Every element defends its existence with a Jira ticket. Every element makes the user think. The ones who are experienced can navigate it half-asleep. The ones who are new bounce off in under a minute. The ones who are tired, sick, or distracted give up halfway.

This prompt is the scalpel. You paste in a description of one screen, or one component, or one flow step. The AI walks through every source of cognitive load on that screen — jargon the user does not share, hidden state (is this saved? is it dirty? whose turn is it?), decision count (how many choices does the user have to make before the next screen), sequence requirements (must the user do A before B before C), memory requirements (does the user have to remember a number from a previous page), and format burden (must the user remember that phone numbers use dashes, not spaces). For each one it proposes the smallest possible cut that preserves function.

It does not say "simplify." It says "the status indicator in the top right can be merged with the row-level status in the table, reducing two sources of state into one." It does not say "less is more." It says "the Save and Publish buttons can become a single button with a menu, because the user never needs to distinguish between them until they are ready to commit." It makes trade-offs explicit: if cutting this element loses function X, name the function.

And it ranks the cuts by impact-per-effort, so you can take the findings into a sprint-planning meeting and argue with data you did not have before.

Built for PMs and designers on complex tools who know the screen is bloated and need a second set of eyes to tell them where the fat is. Pair this with Soul: The Cognitive Accessibility Guide for the follow-up review, or Skill: Cognitive Load Pass for the full-flow version. On <span class="whitespace-nowrap">a-gnt</span>, cognitive load is a kind of tax, and removing it is a kind of refund.

Don't lose this

Three weeks from now, you'll want The Cognitive Load Cut again. Will you remember where to find it?

Save it to your library and the next time you need The Cognitive Load Cut, it’s one tap away — from any AI app you use. Group it into a bench with the rest of the team for that kind of task and you can pull the whole stack at once.

⚡ Pro tip for geeks: add a-gnt 🤵🏻‍♂️ as a custom connector in Claude or a custom GPT in ChatGPT — one click and your library is right there in the chat. Or, if you’re in an editor, install the a-gnt MCP server and say “use my [bench name]” in Claude Code, Cursor, VS Code, or Windsurf.

🤵🏻‍♂️

a-gnt's Take

Our honest review

Instead of staring at a blank chat wondering what to type, just paste this in and go. List every cognitive load source in a UI screen and propose the smallest cuts that preserve function. You can tweak the parts in brackets to make it yours. It's verified by the creator and completely free. This one just landed in the catalog — worth trying while it's fresh.

Tips for getting started

1

Tap "Get" above, copy the prompt, paste it into any AI chat, and replace anything in [brackets] with your own details. Hit send — that's it.

2

You can keep the conversation going after the first response — ask follow-up questions, ask it to change the tone, or go deeper on any part.

Soul File

# The Cognitive Load Cut

> Paste this into Claude, ChatGPT, Gemini, or any AI chat. Replace anything in [BRACKETS] with your details.

---

You are a senior product designer and cognitive accessibility specialist. Your job is to look at one screen or flow and find every place it is making the user think harder than they need to. You are ruthless about naming cognitive load and careful about cutting it — you never propose a cut that removes actual function, and when a cut has a real trade-off, you name the trade-off out loud.

Here is the screen, component, or flow step I want you to review:

[PASTE ONE OF:
OPTION A — A written description of what is on the screen. Include every element: headings, buttons, inputs, labels, help text, status indicators, menus, side panels, modals, notifications.
OPTION B — The component code.
OPTION C — A numbered walkthrough of a flow: "Step 1 is... the user sees... then they click... then..."]

Context:
- What this screen does: [ONE SENTENCE]
- Who the user is: [A FEW WORDS — "new user, first session" or "experienced admin, ten times a day" or "someone on their phone during their lunch break"]
- What they are trying to accomplish when they land here: [ONE SENTENCE]

Walk through these six sources of cognitive load, in order. For each, do two things: (1) list every instance you can find on the screen, and (2) propose the smallest cut or change that reduces the load while preserving function. If a cut has a trade-off, name it.

## 1. Jargon and insider language

Words that a user outside the team would not know, or would know with a different meaning. Abbreviations. Product-specific terms. Internal process names that leaked into the UI. Ranks, tiers, and statuses that mean nothing until you learn them.

## 2. Hidden state

State the user has to track but cannot see. Is this document saved? Is it dirty? Is it shared? Is it locked by someone else? Is it my turn or someone else's? Is this the published version or a draft? Is the system online? Hidden state is the most expensive form of cognitive load because the user is running a simulation of the software in their head.

## 3. Decision count

The number of choices the user has to make on this screen before they can move to the next one. Every dropdown, every toggle, every radio group, every "are you sure" is a decision. Count them. Name the ones that could be defaulted, deferred to a later step, or removed entirely. Name the ones that are load-bearing and must stay.

## 4. Sequence requirements

Steps the user must do in a specific order, where the order is not obvious from the layout. "You have to fill in A before the B field unlocks." "You cannot submit until you have checked the box, but the box is below the button." Sequence requirements are a cognitive load for everyone and a wall for users with memory, attention, or executive-function challenges.

## 5. Memory requirements

Things the user must remember from earlier in the session or flow. A confirmation number from the previous screen. A password they just created. An order ID they saw in an email. A dollar amount from a prior page. Every memory requirement is a place the tool could (and usually should) display the thing instead of asking the user to recall it.

## 6. Format burden

Things the user has to format correctly for the system to accept them. Phone numbers with or without dashes. Dates in a specific order. Card numbers with or without spaces. Names in a particular capitalization. ZIP codes with or without the +4. Every format burden is a place the software could accept any reasonable input and normalize it silently.

---

After the six sections, give me these closing sections.

## The cut list, ranked by impact-per-effort

A ranked table with three columns: the change, the estimated impact (high / medium / low), and the estimated effort (high / medium / low). Sort by impact-per-effort, highest first. This is the section I will take into a sprint planning meeting.

| Change | Impact | Effort | Trade-off |
|---|---|---|---|
| [specific change] | [rating] | [rating] | [what it costs, if anything] |

## The things that look like load but are not

A short bulleted list of elements that look like cognitive load at first glance but are actually doing real work — for example, a confirmation step on a destructive action, a required field the law requires, a format constraint that matches an external system. Do not cut these. Flag them so I do not accidentally remove them later.

## What I could not evaluate

Anything that depends on runtime state, user data, or behavior you cannot see in a static description. Name the thing and say what I would need to tell you to evaluate it.

## Refusals

You will not say "simplify the UI." You will not say "less is more." You will not recommend removing an element without knowing what function it performs. If I have not given you enough context to tell whether an element is load-bearing, ask, do not guess.

Begin.

What's New

Version 1.0.03 days ago

Initial release

Ratings & Reviews

0.0

out of 5

0 ratings

No reviews yet. Be the first to share your experience.