ux
/ux
Szuka, gdzie nowi gracze gubią się lub odchodzą: testy z personami, dane Clarity, recenzje i porównanie onboardingu.
- Autor
- Vamp-pire ↗
- Repozytorium
- hype-me ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 13.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 13.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Szuka, gdzie nowi gracze gubią się lub odchodzą: testy z personami, dane Clarity, recenzje i porównanie onboardingu.
Wyzwalacze z opisu
Argumenty
/ux [url] [quick | deep]
Oryginalny opis (EN)
Find where new players get confused, stuck or leave — first-time playtests by persona on phone and PC (screenshots, time and taps to the first fun moment), real behaviour from Microsoft Clarity (rage/dead clicks, quick backs) and analytics funnels, complaints in reviews and mentions, and a comparison with how similar games onboard — ranked with fixes. Use for "UX 개선점", "어디서 이탈해?", "첫 화면 어때?", "온보딩 봐 줘", "playtest", "why do players leave".
Jak działa
- 0. Missing the game or a URL?
- 1. Gather in parallel (level from the request or profile budget; ranges in references/budg
- 1b. In-game friction (what automatic play cannot feel)
- 2. Judge with game-onboarding principles
- 3. Report
- 4. Next (one line)
Podgląd SKILL.md
---
name: ux
description: Find where new players get confused, stuck or leave — first-time playtests by persona on phone and PC (screenshots, time and taps to the first fun moment), real behaviour from Microsoft Clarity (rage/dead clicks, quick backs) and analytics funnels, complaints in reviews and mentions, and a comparison with how similar games onboard — ranked with fixes. Use for "UX 개선점", "어디서 이탈해?", "첫 화면 어때?", "온보딩 봐 줘", "playtest", "why do players leave".
argument-hint: "[url] [quick | deep]"
---
# Where do new players get stuck?
Answer in the user's language. Every problem needs at least two kinds of evidence, or is marked "single signal".
Follow the shared rules in `${CLAUDE_PLUGIN_ROOT}/references/rules.md` — profile first (run setup, then continue), end with one experiment suggestion, use design checks whenever something visual is involved, and leave a runnable test with anything you make.
## 0. Missing the game or a URL?
Do not stop with only a question. Show the plan in at most 4 lines — level, personas (built from what the user said), which evidence (playtest with screenshots on phone/PC, Clarity, complaints, similar games), and the output (ranked problems, each with evidence and a fix) — then ask for exactly the missing thing (URL or project folder).
## 1. Gather in parallel (level from the request or profile `budget`; ranges in `references/budget.md`)
- **Bot playtests** — start one `ux-playtester` per persona (light: 1, normal: 2, deep: 3). Build personas from the profile's player groups and markets, e.g. "chess beginner, Korean, phone" and "variant fan, English, PC", and give each the goal text (the first fun moment). URL: a local/preview build if available, else the main link. Each agent builds a play bot once (`.hype-me/ux/bots/`) that the browser runs without AI — later runs reuse it, so a check after each update costs almost nothing; the agent only fixes steps that broke.
- **Real behaviour** — a Clarity tool connected in this session, or a Clarity dashboard export in `.hype-me/data/` (dead clicks, quick backs, rage clicks by page) and, if events are tracked, the funnel from the `retention` skill (where the biggest drop is).
- **What players say** — gripes and questions from `.hype-me/mentions.json` / the `mentions` skill ("어렵다", "모르겠다", "느리다", "how do I…").
- **Similar games** (normal and deep) — first screens and onboarding of 1–2 similar games (store screenshots, their sites; Mobbin or getdesign.md for references). Compare: what do they show first, how fast is the first win.
## 1b. In-game friction (what automatic play cannot feel)
Automatic playtests see the first minutes; they do not feel tedium mid-game. Cover it two ways:
- **Friction events** — check whether the game records them (see the `retention` skill's event plan): time per move/turn, undo/back use, the same invalid action repeated, settings changed, menu opened mid-game, game left before the end, rematch/next-game taps. If recorded, show where they spike; if not, propose the 3–5 that fit this game.
- **People test kit** (normal and deep) — prepare `.hype-me/ux/playtest-kit-<date>.md` from `${CLAUDE_PLUGIN_ROOT}/templates/playtest-kit.md`: 3 tasks that go past onboarding into real play, 3 questions, what to watch in Clarity recordings (rage/dead clicks on the board, quick backs), and how to send answers back. The user gives it to 3–5 people; when answers come back, add them as evidence and rerun the ranking.
## 2. Judge with game-onboarding principles
Teach by doing, not reading · first success within ~60 seconds · one obvious thing to do per screen · failing is cheap and you can retry at once · a moment worth showing a friend · works one-handed on a phone (tap targets ≥ 40px, no sideways scroll, readable text).
Mid-game: every tap gets visible feedback (also when it is not allowed — say why) · the next action is obvious after each turn · locked states (waiting, animation, AI thinking) are shown where the player is looking · repeated choices are remembered · undo and restart are one tap away.
## 3. Report
```
UX · <level> · <n> problems
1. <screen> — <what happens to players> · evidence: playtest (persona A stuck 40s), Clarity dead clicks 83% on /app?view=practice, 2 reviews · fix: <concrete change> · effort: small
2. …
Works well: <1–2 things to keep>
Not checked: <what and why>
```
Rank by players affected × how badly it stops them ÷ effort. Save to `.hype-me/ux/report-<YYYY-MM-DD>.md` with screenshot paths.
## 4. Next (one line)
Turn the top 1–2 into experiments (`experiment` skill: hypothesis, metric, check date), or a quick prototype of the fix (`prototype` skill). Bugs found on the way go to the `bugs` list.
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · Vamp-pire/hype-me | 100 | MIT |
| mthines/agent-skills | 100 | MIT |
| Vamp-pire/hype-me | 100 | MIT |
| molveau/skills | 100 | MIT |
| darioabc/massimo | 100 | MIT |
| rainmanjam/poka-yoke | 100 | MIT |
| ericrisco/rsc-harness | 90 | MIT |
| Esposter/Esposter | 90 | Apache-2.0 |
| koolamusic/claudefiles | 90 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 9f3a546417ff (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100heatmap-session-analyst
Rankinguje wskazania Clarity lub Hotjar, takie jak martwe kliknięcia i szybkie powroty, według kosztu w konwersjach GA4. - 100playtest-protocol-designer
Planuje playtesty tak, by dawały użyteczne dane: cele sesji, listy obserwacji, kwestionariusze, wybór typu testu (ślepy, moderowany, A/B), plan zbierania danych i ramy analizy wyników.