Experience Skillsv0.7.1

Skills for coding agents

Better product judgment, not a louder process.

Coding agents can build working software and still make weak product decisions. These portable skills help them see the whole problem and check the result.

  1. 1

    Consider broadly See the whole product problem.

  2. 2

    Intervene selectively Activate only what can materially improve it.

  3. 3

    Verify deeply Check the rendered result, or say what was not checked.

Install Experience Skills

npx skills add moh-obaida/experience-skills

Or let your coding agent install it: paste the prompt into it.

0 Stars0 Forks

Why it exists

Working interfaces are not the hard part.

  1. Centered is not composed

    An empty page with a card in the middle gets called clean.

  2. Known answers get asked again

    A flow repeats questions the software already answered.

  3. The real change is silent

    Motion plays everywhere, and the state that matters updates without notice.

Fifteen skills. Four jobs.

View all skills

Start / Route2

  • experience-architect

    The problem is unclear. Start here.

  • use-all-skills

    A whole product, or an explicit request to consider every skill.

Shape5

  • composition-repair

    Empty, cramped, unbalanced, or only centered.

  • visual-identity

    Interchangeable, or only the logo is recognizable.

  • workflow-compression

    Too many clicks, screens, or repeated inputs.

  • empty-state-design

    Little content, or stuffed with filler.

  • product-friction

    Hard overall: terms, lost context, dead ends.

Behavior4

  • interaction-design

    Controls that are confusing, clever, slow, or unsafe.

  • state-design

    Only the happy path is designed.

  • motion-design

    Motion is generic, missing, or inaccessible.

  • responsive-validation

    Must hold across size, zoom, input, and direction.

Review / Verify4

  • anti-slop-ui

    Generated or template-made, before it ships.

  • anti-ai-slop

    An existing product looks generated.

  • critical-review

    Asked whether an idea is actually good.

  • interface-forensics

    One page needs a full rendered inspection.

The evidence

Three evaluations. Mixed results.

See what was tested and what changed

01 Advice quality

Does installing the skills change the advice?

Principles met out of 51. Without skills 35 and 35. With skills 42, 45 and 45.

Without

35·35/ 51

With

42→45→45/ 51

Improved several scenarios. Activation stayed inconsistent.

02 Exhaustive orchestration

Does using every skill produce a better result?

Not necessarily.

All 14 skills read, but the rendered check never happened.

03 Rendered systems

Do authored design systems stay distinct when rendered?

17 rendered.

Most stayed distinct. Some structures converged. No mobile renders.

Try the current version.

Documented for Codex, Claude Code, and Cursor. This release was tested with Claude Code.

Install Experience Skills

npx skills add moh-obaida/experience-skills

Or let your coding agent install it: paste the prompt into it.

0 Stars0 Forks