interface-design
/interface-design
Decide a user interface's shape before building: states, layout, input paths, and a check plan of run and judged items. USE WHEN designing a UI, screen, TUI or theme. NOT FOR building, proving or reviewing it.
- Author
- Tariq-Anjum ↗
- Repository
- tisserand ↗
- Licence
- MIT
- Last repository change
- 12.11.2026
Assessment and security
- Files in this package
- 2
- Scripts
- 0
- HTML comments
- 1
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
- SKILL.md zawiera komentarze HTML (1). Nie widać ich w podglądzie Markdown, a model je czyta.
What it does and when to use it
Decide a user interface's shape before building: states, layout, input paths, and a check plan of run and judged items. USE WHEN designing a UI, screen, TUI or theme. NOT FOR building, proving or reviewing it.
Original author description
Decide a user interface's shape before building: states, layout, input paths, and a check plan of run and judged items. USE WHEN designing a UI, screen, TUI or theme. NOT FOR building, proving or reviewing it.
How it works
- Quick Index
- Boundaries
- Procedure
- Corrections
- Stop and escalate
- Report
SKILL.md preview
---
name: interface-design
description: "Decide a user interface's shape before building: states, layout, input paths, and a check plan of run and judged items. USE WHEN designing a UI, screen, TUI or theme. NOT FOR building, proving or reviewing it."
disable-model-invocation: false
---
# Interface Design
## Quick Index
- **Output:** a brief (surface, path, layout, states) and a check plan.
- **Split:** each check is RUN (executed or inspected) or JUDGED (criteria, decider).
- **Fail-closed:** an open required check is never ready.
- **Generic:** no platform or app-type depth; project conventions win.
## Boundaries
Not this skill when building it (implement), proving your own built change
(verify), judging finished work (review), shaping code, data or services
without a user surface (other design), settling goal or done (clarify), or
ordering the build (plan). Authorship does not change the type:
redesigning your own or another's surface is design; checking a built one
is verify for your change, else review. The brief's author never reviews
it.
## Procedure
1. **Surface.** Name it (web page, app screen, terminal UI, widget) and its
inputs (pointer, touch, keyboard, screen reader). If an unknown surface
or user goal would change the design, halt and ask.
2. **Inventory.** Read only: tokens, components, platform conventions,
project rules. Reuse first; list reused and new parts.
3. **Brief.** Main task path; layout at small, normal and large sizes;
element states (default, focus, hover, active, disabled, loading, empty,
error, success); long text; motion, reduced form; values per theme.
4. **Bar.** The project's named accessibility conformance level; with none,
propose one and ask: until the owner confirms it, checks it sets stay
open. Stricter levels and spacing grids only where adopted.
5. **Check plan.** Per requirement: method, pass threshold, failure
observation; required unless its source says optional. RUN only with a
named tool, script or inspection step; otherwise JUDGED. Candidates:
- contrast of text and controls, computed from token values;
- theme parity: each token has a value per theme, each pair passes;
- keyboard path: a scripted key walk operates every action, focus order
follows reading order, focus stays visible;
- reduced motion: with the setting on, only motion marked essential runs;
- no clipping or overflow at the named window or terminal sizes;
- web or app controls expose a name and role; a terminal UI restores
the terminal on exit.
JUDGED checks (hierarchy, clarity, convention fit, copy) need written
criteria and a decider. A check that is neither, or a RUN probe still
failing to run when the budget ends, is UNTESTABLE.
6. **Baseline,** only if a surface exists: run build-free RUN checks (e.g.
contrast from a token file) as scratch probes; keep outputs; change
nothing tracked.
7. **Status,** first match:
- a requirement conflicts with an owner or project rule, or needs
authority you lack: STOP;
- a required check is UNTESTABLE, rests on an unconfirmed bar or
assumption, lacks an answer, or a baseline failure is left
unaddressed: OPEN;
- every required check has method, threshold and failure observation,
or criteria and decider: READY_FOR_BUILD;
- otherwise: OPEN.
READY_FOR_BUILD is not proof. The build takes the plan and the probe
correction count (continued in the same task); the result is verified
against the plan, reviewed where gated. Never claim compliance.
## Corrections
Repairs touch only your probes: first run attempt zero, then at most three
corrections per task under the correction ladder
(`policy/correction-ladder.md`): no reset, count
stated in handoffs; side effects inspected (section 10); OBSERVATION to
NEXT_DECISION (section 11), changed method; research when uncertain, fresh
before correction 2, deep before 3; MEDIUM fields first; checkpoint before
HIGH. Never rerun an unchanged probe.
## Stop and escalate
Ask the owner before real user data, credentials, production, a brand or
legal claim, or work outside the gate; when you cannot demonstrate that a
repair is below CRITICAL; after three failed corrections (reassess, read-only).
## Report
Brief; check table (kind, method, threshold, failure sign, state);
baseline; assumptions; status; correction count; open items.
<!-- provenance: original text (2026-11-08, rev 2026-11-09). Mechanisms
only: ifd-01 (no platform packs), ifd-02, ifd-03, ifd-04 (licence
unverified). Owner rules: Engineering Workflow, correction policy. -->
Files in this package (2)
- SKILL.md 5 KB
- capability.toml 2 KB
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · Tariq-Anjum/tisserand | 90 | MIT |
| reactive/data-client | 100 | Apache-2.0 |
| zpyoung/quirk | 100 | MIT |
| tushaarmehtaa/tushar-skills | 100 | MIT |
| nttare0/all- | 100 | MIT |
| 3xC1ibR/interface-design | 100 | MIT |
| dammyjay93/interface-design | 100 | MIT |
| seikaikyo/dash-skills | 90 | MIT |
| hamzzaqureshi/gravidex | 90 | MIT |
Database snapshot: 13.11.2026 · package version 0c7f00f0fb4e (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100web-design-guidelines
Set a product-specific design direction before building web UI, then build and review against concrete interface rules - interactions, forms, content states, typography, layout, motion, theming, performance - with a rendered finish check and terse file:line findings. Use when… - 100design-build
Implement or refine frontend visual design, CSS, layout, typography, color tokens, and component states. Use when creating or styling an interface. - 90interface-designer
Use when a product flow needs interface structure, component planning, interaction details, or screen-level UX shaping before implementation. - 90interface-designer
Use when a product flow needs interface structure, component planning, interaction details, or screen-level UX shaping before implementation. - 90frontend-design
Give a web page or app its own look instead of a generic AI-made one: choose colors, type and layout on purpose before building, then critique the result. Use when building a new interface or reshaping an existing one. - 90ui-copy
Write or review interface text such as labels, buttons, headings, field hints, empty states, errors, confirmations, tooltips, and onboarding messages. Use when building or changing a screen, component, or flow that shows text to users, even for a single label or message, and…