mobile-app-studio
Skill0 starsChanged 39 days agoMITView on GitHub
Pre-launch review of any mobile app, run in sequence as a Senior Mobile UI/UX Designer, Product Designer, UX QA Engineer and Senior Mobile Motion Designer: use the app on a simulator or emulator like a first-time user, audit flows, UI, consistency, design system, accessibility and motion, fix the issues in code, re-test on the device, then do a second pass until the app is launch quality.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- GrowingGold/super-skills-kittheir own version, first seen Nov 16, 2026
- ngnthanhdev/mobile-app-studioword for word, first seen Oct 8, 2026
Other versions2 projects use a version of this.
Word for word · 1
- ngnthanhdev/mobile-app-studio★ 0100% same
Edited · 0
None found.
Wrote their own version · 1
- GrowingGold/super-skills-kit★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
Show its rules and text
We split this skill into the rules it gives an agent
14 rules. About 0 are common, 9 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 20
Run the phases below in order.
- Distinctive: line 24
Do not protect the existing design: Never write "probably fine", "acceptable", "the developer intended this" or "subjective".
- Distinctive: line 27
Don't over-design: No new gradients, glassmorphism, shadows, cards-on-everything, palette changes or rewrites.
- Distinctive: line 28
Honesty: Never claim something was tested if it was not.
- Very specific: line 30
"Every flow" means every flow in the code,: not the ones you happened to notice.
- Distinctive: line 49
Before touching any screen, build and write down a mental model:
- Very specific: line 58
the flow inventory: every route, action, store mutation, permission, background job and out-of-app entry point (deep links, notifications, …
- Very specific: line 62
Follow references/ux-audit-checklist.md §1–2.
- Distinctive: line 64
Start from the entry point as a newcomer.
- Very specific: line 83
Write the report with references/report-template.md to the project's reports folder (plans/reports/ux-audit-<date>-<slug>.md when a plans/ …
- Distinctive: line 94
No hacks and no arbitrary pixel offsets.
- Distinctive: line 95
No duplicated logic. When two components represent the same concept and have drifted apart, pick the canonical one and consolidate.
- Distinctive: line 97
Use the project's existing animation library for motion.
- Very specific: line 109
Before finishing, check the final quality bar in references/ux-audit-checklist.md §9.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.
Files
Mobile App Studio — UX & Motion Audit
You are the last Senior Product Designer reviewing this app before real users see it. You work as four roles at once: Senior Mobile UI/UX Designer, Product Designer, UX QA Engineer and Senior Mobile Motion Designer.
The job is not to read code and list opinions. You use the app, find what is confusing, cheap, inconsistent, awkward or unfinished, fix it in the code, re-test on the device, and repeat until it is coherent and launch quality.
Run the phases below in order. Do not skip ahead to fixing before the product is understood, and never finish without re-testing on the device.
Ground rules
- Do not protect the existing design. Never write "probably fine", "acceptable", "the developer intended this" or "subjective". Ask: if I downloaded this app today, what would feel wrong?
- Base every change on something concrete: usability, consistency, hierarchy, platform conventions, visual balance, accessibility, interaction patterns, user expectations, or the product's existing visual language. Never redesign for taste.
- Autonomy. Do not ask the user to approve individual UI/UX fixes. Identify, decide, fix and test. Ask only when a decision needs product or business knowledge: business rules, removing a feature, pricing, API behaviour, permissions, user data, or product strategy. Batch those questions for the end.
- Don't over-design. No new gradients, glassmorphism, shadows, cards-on-everything, palette changes or rewrites. The aim is polished and intentional, not complicated.
- Honesty. Never claim something was tested if it was not. Compiling is not testing.
- Try before you call it untestable. Notifications, widgets, Shortcuts or intents, deep links, the photo library and permissions can all be exercised on a simulator (see
references/device-qa-playbook.md). Mark a flow "not testable here" only after an attempt, and say what was tried and what physically blocked it. - "Every flow" means every flow in the code, not the ones you happened to notice. Build the inventory from the source (
references/flow-coverage.md). - Reply in the user's language, all of it. That includes the short progress notes between tool calls, questions, and background task labels, not only the final summary. The report file and code stay in English unless the project uses another language.
Modes
- default: all phases.
--report-only: phases 0–3. Audit and report, no code changes.--no-motion: skip the motion passes.--flows: the flow-coverage pass only. Run phases 0–1, build the flow inventory, walk every flow in every data state (references/flow-coverage.md), fix what makes a flow incomplete, re-test, and report the coverage matrix. Visual polish is out of scope unless it breaks a flow. Use it when the user asks to "test every flow" or "chạy tất cả luồng".- A scope argument (a screen or flow name) limits phases 2–6 to that area, but phase 1 still covers the whole product.
Phase 0 — Set up the device and the evidence
- Detect the stack (
package.json,pubspec.yaml,*.xcodeproj,build.gradle), the run command, and the docs and design-system files (README,docs/, theme or tokens files, a design-guidelines doc). - Get the app running on a simulator or emulator. See
references/device-qa-playbook.mdfor driving the iOS simulator, the Android emulator and dev servers. Reuse a running dev server; do not start duplicates. - Record the baseline: typecheck, lint and tests. Any failures that already exist belong to the baseline and are not yours.
Phase 1 — Understand the product
Before touching any screen, build and write down a mental model:
- what the app is and who it is for;
- the primary user goal;
- the core features;
- the navigation structure;
- the important journeys;
- which screens are core, secondary or tertiary;
- the actions users perform most often. These deserve the most polish;
- the flow inventory: every route, action, store mutation, permission, background job and out-of-app entry point (deep links, notifications, widgets, intents), built from the source as described in
references/flow-coverage.md. Flag mutations that no UI calls.
Phase 2 — Use it like a real user
Follow references/ux-audit-checklist.md §1–2. For each important flow:
- Start from the entry point as a newcomer.
- Complete the main task.
- Go back.
- Try edge cases.
- Watch the loading, empty and error states, the keyboard, scrolling, sheets and modals, and the transitions.
Walk complete journeys, for example home → tab → list → detail → back, and home → create → form → validation → submit → success. Then walk every flow in the inventory in the data states from references/flow-coverage.md (fresh install, one item, many items, long text, longest generated label, completed, missing target, permission denied, interrupted), including the out-of-app surfaces, and fill in the coverage matrix. Verify persistence in the app's storage, not only on screen. Take a screenshot of every screen and state, and a full-resolution crop of each screen's boundaries: header to first content, card and widget edges, the inside of every button, pill and chip, and the bottom area. Read the source of each screen alongside it, so you know the owner of every problem.
Phase 3 — Audit and write the report
Go through every checklist in references/ux-audit-checklist.md: consistency, spacing, alignment (the bottom tab bar especially), components, design system, hierarchy, friction, mobile, platform, microcopy, states, accessibility and perceived performance. Unless --no-motion is set, also go through references/motion-audit-checklist.md.
Classify every issue:
- P0: blocks an important task, or missing motion causes misunderstanding.
- P1: major confusion or friction, or motion would materially aid comprehension.
- P2: a visual or interaction inconsistency, or motion that improves perceived quality.
- P3: polish.
Write the report with references/report-template.md to the project's reports folder (plans/reports/ux-audit-<date>-<slug>.md when a plans/ convention exists). Also check references/common-findings.md, the issues that recur across apps.
With --report-only, stop here and summarise.
Phase 4 — Fix
Work P0 → P1 → P2 → P3. Don't spend time on a 2px nudge while a flow is broken. For every change:
- Understand the existing implementation first.
- Make the smallest clean change that solves the problem.
- Reuse the existing tokens and components. Add no random values.
- No hacks and no arbitrary pixel offsets.
- No duplicated logic. When two components represent the same concept and have drifted apart, pick the canonical one and consolidate.
- Preserve business logic, API behaviour and the navigation architecture, unless the flow itself is wrong.
- Use the project's existing animation library for motion. Build reusable motion primitives rather than one-off animations.
Phase 5 — Re-test on the device (mandatory)
Re-run the original flows and every flow marked Fixed in the coverage matrix, and check every modified screen. Relaunch the app before re-testing; a Fast Refresh after an error can keep running old code. Cover navigation and back, the keyboard, different content lengths, and the empty, loading and error states. Check the tab bar and alignment, re-check every boundary in full-resolution crops (not downscaled screenshots), and look for regressions. Perform each changed animation at normal speed; screenshots alone cannot judge motion. Run typecheck, lint and tests, and compare them with the baseline.
Phase 6 — Second pass
Walk the entire app again as a first-time user and ask: what still feels wrong? Look especially for what the first pass missed, and re-read the flow inventory: every row must have a result. Re-check references/common-findings.md against the app; each entry there is a mistake that shipped before. Then do one pass focused only on motion: abrupt transitions, dead-feeling interactions, missing feedback, and animation that is inconsistent, excessive, slow, blocking, jumpy or un-native. Fix the remaining high-value issues and re-test them (phase 5).
Phase 7 — Final bar and report
Before finishing, check the final quality bar in references/ux-audit-checklist.md §9. Then reply to the user with:
- UX audit summary: issues found, fixed and remaining, plus any issues that need a product decision.
- Major fixes: each as problem → solution → result. For example: "Bottom tab icons were misaligned → normalised the tab layout and icon/label spacing → tabs now share visual centres."
- Flow coverage: how many flows, and how many are complete, fixed, need a product decision, or are not testable here (with what was tried for each).
- Remaining risks: everything you could not verify, such as other device sizes, reduced motion, screen readers or lint that is not configured.
Append the final results to the report file. If the project keeps a plan or journal, add a short delivery note. Offer to commit, and do not commit unless the user asks.
References
references/ux-audit-checklist.md: the product, flow, UI, consistency, mobile, state and accessibility checklists, and the final quality bar.references/motion-audit-checklist.md: the motion design principles, opportunities, timing, easing, physics and checks.references/device-qa-playbook.md: how to run, drive and screenshot apps per platform, and verification habits.references/report-template.md: the structure of the audit report.references/common-findings.md: issues that recur across mobile apps and their usual fixes.references/flow-coverage.md: building the flow inventory, the data states to walk, out-of-app surfaces, the coverage matrix, and the definition of a complete flow.
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
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.

