SkillAgentSearch skills...

ux

Your agent invents things because it has no memory, and can't touch your database because it has no arms. rsc is the meta-harness that gives it both, plus the trade to know the job and the reflexes to not break your repo. For Claude Code, Codex and 15 more.

Install / Use

npx skills add ericrisco/rsc-harness --skill ux

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

72/100

Supported Platforms

Universal

Our assessment of ux

ux scores 72/100 on our quality scale, 552nd of 604 Data & Analytics skills we index.

Its SKILL.md is 9.2 KB long, well organised into 10 sections and no code examples: a thorough specification that gives an agent plenty to work with.

It has no GitHub stars yet, so there is no community track record; judge it on its content.

Substance
29/30
Structure
13/20
Description
15/15
Adoption
0/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated yesterday, so ux is actively maintained.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 80/100, with 2 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

ux compared with similar skills

All 4 of these similar skills score higher than ux; compare them before choosing.

SkillScoreStarsUpdatedFormat
ux (this skill)by ericrisco7201d agoSKILL.md
taste-skillby Leonxlnx100116.8k8d agoCLAUDE.md
Claude-Code-Game-Studiosby Donchitos10039.7k7d agoCLAUDE.md
x402by browser-use100121.0k3d agoSKILL.md
algorithmic-artby anthropics100177.9k55d agoSKILL.md

Frequently asked questions

How do I install ux?
Run npx skills add ericrisco/rsc-harness --skill ux. The install tabs above show the steps for each supported agent.
Which AI agents does ux work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is ux safe to use?
It declares no license and scores 80/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is ux still maintained?
The repository was last updated yesterday, so ux is actively maintained.

name: ux description: "Use when an app or SaaS must be easy to use: designing complete user flows, screens and every state, navigation, forms, onboarding, tables, dashboards, search or AI features, or auditing usability and accessibility. Anyone finishes the job without a manual; experts go fast. NOT the visual direction (design), NOT the words (landing-copy)." tags: [ux, usability, user-flows, flows, screens, states, empty-state, signup, modal, saas, app, product-design, navigation, forms, onboarding, dashboard, productivity, accessibility, wcag, interaction-design, usabilidad, flujos, pantallas] recommends: [design, accessibility, landing-copy, react] profiles: [minimal, core, full] origin: risco

ux

The bar: someone who has never seen the product finishes the main job without asking anyone, someone who uses it every day finishes it fast, and nobody loses work when something fails. A screen that looks good and misses that bar is not done.

  • The flow comes before the screen. Map the whole journey, from arrival to the job done and back again, before drawing anything.
  • Every screen answers five questions: where am I, what is this, what can I do, what just happened, where do I go next.
  • Every state is designed, not only the full one: empty, loading, error, offline, no permission, too much data.
  • You prove it by using it, in a real browser, with the keyboard, at phone width. Reading the code is not a check.

The depth ships in upstream/, two MIT skills copied unmodified (THIRD_PARTY.md). This file is the method; load only the reference the task touches (table in Files).

0. Name the job

Write four lines before anything else: who (first-time, occasional, daily expert, admin), the job in their words, done when (what they can see), and the risk (money, data, legal, irreversible). Assume motivated but distracted people with average domain knowledge and imperfect memory.

Ask one question only when the answer changes the design: the main task is unclear, the action is risky or irreversible, novices and experts need different things, the platform is unknown, or the request asks for a hidden or manipulative pattern. The ready-made questions, each with its default, are in upstream/ux-usability-foundations/references/decision-prompts.md. Everything else is decided by the defaults in upstream/ux-usability-foundations/GUIDE.md.

1. Map the whole flow

Use references/flows.md: the flow map, the lifecycle (sign-up, empty account, first value, the daily loop, limits, permissions, failures, offline, leaving and coming back), the second role (whoever is invited, not only the admin who set it up), and the step count of the core loop. Output a table with one row per step and screen. Missing lifecycle rows are the usual reason an app "feels broken".

2. Design each screen

  1. One job and one primary action. The primary action is named with a verb and an object ("Send invoice"), it is the strongest element, and the destructive action is never the primary one.
  2. Pick the pattern from the job: list, detail, form, wizard, settings, dashboard, search, or an overlay. Decision trees (modal, side panel or page; which notification type) are in upstream/ux-designer/GUIDE.md. Navigation and IA in 05-information-architecture.md.
  3. Fill the state matrix in references/flows.md for every screen: default, loading, empty (new account and filtered to zero), partial, error, offline, no permission, overflow, pending.
  4. Prevent, then recover: constraints and safe defaults first, inline validation, input kept across every failure, undo for anything reversible, confirmation only for the irreversible. Forms in 07-forms-and-inputs.md.
  5. Speed for experts: keyboard shortcuts, command search, bulk actions, saved views, remembered filters, sensible defaults. They never replace the visible route for everyone else.
  6. Accessible from the start: native elements first, keyboard reaches everything, visible focus, labels, colour never the only signal, targets of at least 24×24 CSS px. WCAG 2.2 AA in 03-accessibility.md.

The rules with their numbers (contrast, type, timing, targets, toasts) are in upstream/ux-designer/GUIDE.md › Core rules and Reference values. Use them; do not invent others.

3. How it looks

The visual direction (type, palette, composition, pictures) is ../design/SKILL.md's, and it runs its engine on the screens this skill has specified. What stays here is the hierarchy that serves the job: one dominant element per screen, three levels of emphasis, one spacing scale, colour that means state, density chosen by the job (monitoring dense, onboarding calm). The tells of generated interfaces are listed in upstream/ux-designer/GUIDE.md › Generated-UI defaults to avoid, and visual detail in 04-visual-design.md.

4. Prove it by using it

Walk the main job from a cold start in a real browser (the Chrome DevTools tools, when present): a new account, then one with data. Then the extremes: keyboard only, 320px wide, 200% zoom, a long translated string, empty data, a lot of data, a slow network, a failed request. Run axe or Lighthouse when the browser is there. Write down every hesitation and fix the hesitations before polishing anything. A check that could not run is reported as not run.

Review mode

When the task is to audit an existing interface: steps 0 and 4 first, then report findings ranked by impact on the job, in the format of upstream/ux-designer/GUIDE.md › Reviewing or auditing a UI (Blocker / Major / Minor, where, fix, reference). Inspection order and the full checklists: upstream/ux-usability-foundations/GUIDE.md › Workflow for critiquing and references/checklists.md. Split quick wins from structural changes. If nothing serious turns up, say so in one line.

Report

Say briefly: the person and the job; the flow map; the screens with their primary action and the states covered; what was walked in the browser and what was not; the findings or decisions with their reference; what still needs real data or real users to confirm.

Hand-offs

  • Visual direction, type, palette, pictures, landing pages → ../design/SKILL.md.
  • The words on a marketing or conversion page → ../landing-copy/SKILL.md.
  • A formal WCAG conformance programme or audit → ../accessibility/SKILL.md.
  • Component code, state and data fetching → ../react/SKILL.md or ../nextjs/SKILL.md.
  • Measuring a flow with real users at scale → ../analytics/SKILL.md, ../ab-testing/SKILL.md.

Files

| Load when the task involves… | File | | --- | --- | | The whole journey, lifecycle, roles, state matrix | references/flows.md | | Workflows, core rules, numbers, decision trees, generated-UI tells | upstream/ux-designer/GUIDE.md | | Principles, defaults, questions to ask, critique order | upstream/ux-usability-foundations/GUIDE.md | | Questions with their recommended default | upstream/ux-usability-foundations/references/decision-prompts.md | | Principles as cards, full checklists, anti-patterns | upstream/ux-usability-foundations/references/ › principle-cards.md, checklists.md, anti-patterns.md | | Nielsen's heuristics, Gestalt | upstream/ux-designer/references/01-core-principles.md | | Fitts, Hick, Jakob, Tesler, peak-end | …/02-laws-of-ux.md | | WCAG 2.2, ARIA, EAA | …/03-accessibility.md | | Type, colour, spacing, hierarchy, dark mode | …/04-visual-design.md | | Navigation, sitemaps, card sorting | …/05-information-architecture.md | | Modals, tooltips, drag and drop, motion timing | …/06-interaction-design.md | | Forms, validation, input types, errors | …/07-forms-and-inputs.md | | Touch, gestures, responsive, mobile navigation | …/08-mobile-ux.md | | Microcopy, error messages, tone | …/09-ux-writing.md | | Interviews, usability tests, surveys, metrics | …/10-user-research.md | | Tokens, component APIs, design system docs | …/11-design-systems.md | | Presence, cursors, avatars | …/12a-presence-awareness.md | | Conflicts, sync, offline, sharing, history | …/12b-conflict-resolution-sync.md | | Canvas zoom, pan, selection; layers, snapping, performance | …/13a-canvas-navigation.md, …/13b-canvas-objects-performance.md | | Chat, copilots, agents, generative UI | …/14-ai-ux-patterns.md | | Dark patterns, consent, DSA and GDPR interface rules | …/15-ethical-design.md | | First run, activation, empty states, checklists | …/16-onboarding.md | | Notifications, push, toasts, preferences | …/17-notifications.md | | Charts and dashboards | …/18-data-visualization.md | | Search, autocomplete, filters, zero results | …/19-search-ux.md | | Trust, delight, recovering from errors | …/20-emotional-design.md | | Tables, sorting, pagination, bulk actions | …/21-data-tables.md | | Loading, skeletons, optimistic updates, Core Web Vitals | …/22-performance-ux.md | | Localisation, RTL, Intl, plurals | …/23-internationalization.md | | Voice and multimodal input | …/24-voice-and-multimodal.md | | Barcodes and QR: scanning, scanner workflows, display | …/25a-barcode-camera-scanning.md, …/25b-barcode-scanner-workflows.md, …/25c-barcode-display.md | | Which upstream versions ship here, with file hashes | upstream/UPSTREAM.json |

…/ is upstream/ux-designer/references/.

Related Skills

View on GitHub
GitHub Stars0
CategoryData
Updated1d ago
Forks0

Trust signals

80/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 medium1 low