Workshop 01 — Discovery

Time: 60 minutes solo, 90 minutes for a pair.

Goal: Run a real discovery pass on a real screen. End with a categorized opportunity map you’d be willing to put in a PR description.

Setup (5 min)

Open your real product. Pick one non-trivial screen — a dashboard, a settings page, a primary list view. Don’t pick a marketing page.
Open Claude Code in the repo: cd ~/your-project && claude.
Have 03-ui-ux-vocabulary/full-cheat-sheet.md open in another window.

Step 1 — Run pattern discovery (15 min)

Paste Prompt 01 — UI Pattern Discovery into Claude Code, with your screen as the target.

Read the entire response. Don’t skim.

Step 2 — Categorize and triage (15 min)

For each suggestion in the response, write next to it:

Impact: High / Medium / Low (how much does this help users?)
Effort: Low / Medium / High (how long to ship?)
Risk: Low / Medium / High (chance of regression / wasted work?)

Then sort by impact-per-effort.

Step 3 — Pick 3 to pursue (10 min)

The top 3 impact-per-effort items are your candidates. Write them up:

## Top 3 candidates from discovery

1. [Pattern name] — [why it helps users] — [effort estimate]
2. [Pattern name] — [why it helps users] — [effort estimate]
3. [Pattern name] — [why it helps users] — [effort estimate]

Step 4 — Vocabulary check (10 min)

Look at the 3 patterns you picked. For each, can you:

Name the pattern in one sentence?
Name 2 alternative patterns and why they’re worse for this use case?
Name a real component in your stack you’d use to implement it?

If any answer is “no,” open the relevant page in 03-ui-ux-vocabulary and read until you can answer all three.

Step 5 — Write the PR description (5 min)

Pretend you’re filing a PR for the #1 candidate. Write the description:

## What
[1-2 sentences: the change]

## Why
[1-2 sentences: the user problem]

## Pattern
[Pattern name and link to the cheat-sheet entry]

## Components reused
[Specific files in the repo]

## Verification
[typecheck / lint / responsive / a11y / screenshots]

Time-box this to 5 minutes. The act of writing is the exercise.

Self-grading rubric

Score yourself out of 10 on each:

Pattern naming. Can you name every pattern in the response, or are some still unfamiliar? (10 = all named; 5 = half; 0 = couldn’t repeat any)
Triage discipline. Did you actually score impact/effort/risk for every item, or did you skim and pick what felt good? (10 = all scored; 0 = picked from gut)
Tradeoff articulation. For your #1 pick, can you explain in one paragraph why two alternatives are worse? (10 = clear; 0 = couldn’t)
Reuse vs. invent. Did you identify which components in your repo would be reused? (10 = yes; 0 = no)
PR-readiness. Could a teammate read your PR description and approve it without followups? (10 = yes; 0 = vague)

A score of 35/50 means you’re functional. 45/50 means you’re ready to pair-teach this.

Common failure modes

You skimmed the response. The response is the exercise. Read it slowly.
You picked by gut instead of triaging. Force yourself through the impact/effort/risk grid even when the answer feels obvious. The discipline is the value.
You couldn’t name alternatives. That’s a vocabulary gap. Mark it. Read the relevant cheat sheet entry.
You skipped the PR description. Writing what you’d do is the difference between a workshop and a daydream.