Clarify the brief BEFORE any design work to avoid building the wrong thing. Interrogates goal, user, constraints, success criteria, and scope, then writes a short context doc the other design skills consume.
SkillScore91Maintained
Automatic
◆Design & UX Review pack
Last verified 8/6/2026
What it changes
Passed 20 of 22 test cases with the skill, and 14 of 22 without it.
Without skill64%
With skill91%
Checks it passed that the model alone missed
The response creates or references the specific path `./design/context.md`.
The output limits the initial question set to concise, high-priority unknowns.
The generated file groups user behavior hypotheses into an open assumptions section.
and 4 more in the full report
7 fixed13 passed either way1 still failing1 broke
Pass-rate lift
+27 pp
Tokens
-2%
Agent turns
0%
Response time
-34%
Model gemini-3.6-flashMeasured 8/6/2026v120 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 7
Prompt
We already have a fully validated `./design/context.md` defining goals, constraints, non-goals, and risks for our onboarding redesign. Please run the discovery questioning session again to ensure nothing was missed.
What was checked, without and with the skill
Checked
Without skill
With skill
The response declines to re-interview or restart context discovery when valid context already exists.
Missed
Passed
case-13 · gemini-3.6-flash · 8/6/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md61 linessingle file
Design — Get Context
De-risk the whole design effort by nailing the brief first. The cheapest fix is the question you ask before building. This skill produces ./design/context.md — the single source of truth every other design skill reads. Skipping it is the #1 cause of "that's not what I meant" rework.
When to use
Any new design/UI task where the spec isn't already crisp.
The request is a one-liner ("design a settings page") with implicit, unstated requirements.
Before design-research, design-audit, design-ideate, or any build skill.
When NOT to use
A solid ./design/context.md (or equivalent spec) already exists → proceed to the next skill.
A pure visual-asset task with no UX → that's the Creative Production set.
What to interrogate (ask the gaps, don't dump all at once)
Goal: what outcome or behavior change is this UI for? (not "a page" — why)
User: who uses it, their context, tech comfort, frequency.
Job-to-be-done: the core task the screen must make effortless.
Success criteria: how we'll know it worked — ideally measurable.
Scope: must-have vs nice-to-have for v1; explicit non-goals.
Ask only the questions whose answers would change the design. If something's already given, don't re-ask. If the user can't answer "success criteria", propose 2-3 candidates and let them pick.
Workflow
Identify the critical unknowns (the ones that would change what you build).
Ask them concisely, grouped, not as a 20-question interrogation.
If the user's framing seems flawed (wrong problem, contradictory constraints), raise it and propose an alternative before proceeding.
Write ./design/context.md with answers + explicitly flagged assumptions + risks.
Hand off to design-research, design-audit, or design-ideate.
Output: ./design/context.md
# Context
goal: cut onboarding drop-off (currently 40% abandon at step 2)
user: first-time SaaS trial users, mixed tech comfort, on desktop
JTBD: get to "first value" (a created project) in under 2 minutes
constraints: stack=Next.js + Tailwind DS=internal "Aurora" platform=web deadline=2wk
success: step-2 completion ↑ from 60% → 80%; time-to-first-project < 2min
scope v1: [3-step wizard, skip-able, progress saved]
non-goals: [mobile layout, SSO, team invites]