agentleFS
ExagrowExagrow/dashboard-workshop/.claude/skills/jobs-quote-ux/SKILL.md

jobs-quote-ux

Skill0 starsChanged 39 days agoMITView on GitHub

Test a design against the user-experience-first standard, naming the person first and working backward from what they are trying to accomplish to the technology. Use when someone asks whether a flow, screen, dashboard, skill, process or error message is actually any good for the people who use it, or says "apply the Jobs standard", "work backwards from the user", "critique this UX", "is this the right experience", "does this pass". With no target named, it reports the quote and stops.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not ask for one and do not choose one yourself.

    line 19
  2. Distinctive:

    Every control says what will happen and afterwards says what did.

    line 55

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

The user-experience-first standard

Steve Jobs, closing Q&A at Apple's 1997 developer conference:

"You've got to start with the [user] experience and work backwards to the technology."

He went on to say you cannot start with the technology and then go looking for where to sell it, and that he had made that mistake more than anyone in the room and had the scar tissue to prove it.

If no target was named

Do not ask for one and do not choose one yourself. Report the quote and the context above, then stop. Everything below applies only when there is something to test.

How to apply it

Name the person first, specifically. Not "the user". A finance analyst upskilling into dashboards. A VP looking at a number that is wrong. A dashboard owner clearing a queue on a Tuesday afternoon. The design changes depending on which, so pick.

Say what they are actually trying to accomplish, in their words rather than the system's. A person manages notifications, not webhook config. A person wants the number to be right again, not a surgical revert.

Then work backward. From that need to the interface, from the interface to the mechanism, and only then to what has to be built. If a step exists because of how the system is put together rather than because the person needs it, that step is technology leaking upward.

Then say where the current design fails the test, concretely:

  • What does it ask the person to know that is really about the implementation? Branches, refs, pull requests, versions and promotions are all suspects.
  • What does it ask them to decide that the system could decide, or that has only one sensible answer?
  • Where does it report the mechanism rather than the outcome? "409 conflict" instead of "this cannot ship yet, and here is why."
  • What is on screen because it was easy to render rather than because it is needed?

Cut to the essential. Dumping everything available and calling it complete is the failure mode this standard exists to prevent. Fewer, better-chosen things beat exhaustive ones.

What good looks like

The person accomplishes the thing without learning the system's shape. Where the system's shape is genuinely unavoidable, it is named in their language and once, not repeatedly. Every control says what will happen and afterwards says what did.

Where this sits

This standard comes first in the set, and it sets the target for the others.

  • ux-heuristics is the systematic pass. Once the person and their need are named, walk the real screen against Nielsen's ten heuristics to find where it fails them.
  • closed-loop-visual-feedback supplies the evidence. Judge the rendered thing, at the size and on the device the person will use, never the source.
  • inverse-set-kondo asks the same "does this earn its place" question of what sits behind the screen: rules, checks, config, and code that have piled up.

Output

Say what the person needs, what the design currently makes them do instead, and the smallest change that closes the gap. If the honest answer is that the design already passes, say so plainly and stop; inventing a critique to fill the template is its own failure of the standard.


Created by Exagrow AI Consulting. Provided under the MIT License.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.