Build enterprise SaaS application surfaces — data tables, list views, dashboards, settings, admin panels, detail/inspector panels, multi-step flows, command palettes — that read as designed rather than assembled. Use when building or redesigning any in-app product surface, CRUD screen, internal tool or admin UI; when a table needs columns, sorting, filters, bulk actions, pagination or empty states; when a form needs validation timing, progressive disclosure or a destructive-action pattern; when an app screen feels cluttered, cramped, "too much chrome", inconsistent, or like a marketing page wearing an app costume. A capability catalog of grounded UX/UI decision rules with real numbers — chrome budget, app type scale, 4px spacing, control tiers and hit targets, table anatomy, product-UI motion, WCAG 2.2 AA — plus drop-in React + Tailwind v4 reference implementations. Do not use for marketing pages, landing pages, or visual-style exploration.
Select a color palette and type pairing that fits a brand, audience, and product, and prove it works. Defines palette roles (neutral ramp, brand, accent, semantic states, surfaces), checks every intended foreground and background pair against WCAG contrast thresholds, plans dark mode and color-blind-safe signaling, chooses font roles and pairings by shared structure and clear contrast, verifies script coverage, numerals, weights, licensing, and performance, and sets a type scale. Use when someone asks for a color palette, a font pairing, typography or brand color choices for a site or app, whether two fonts go together, or whether a palette is accessible. Do not use for exploring whole visual styles (use explore-visual-directions), building a moodboard (use creative-moodboard), auditing existing screens (use design-audit), or governing a shared component library (use component-library-governance).
Selects color palettes and type pairings based on brand, accessibility, audience, and product context.
Define how a design-system component library is owned, extended, reviewed, versioned, and retired. Sets the ownership model and decision rights, an intake and triage path for requests, component status levels with entry and exit criteria, a definition of done covering design, accessibility, code, and documentation, a scored review rubric, versioning and deprecation rules, exception handling, and health metrics with formulas. Use when someone asks for design system governance, contribution rules, component status levels, a review checklist for new components, how to deprecate a component, or why teams keep forking or detaching components. Do not use for auditing screens for inconsistency (use design-audit), choosing colors and fonts (use color-font-pairing), testing built components for accessibility (use accessibility-qa), or app-surface patterns like tables (use steel-design).
Defines design-system component governance, contribution rules, status levels, and review criteria.
Audit an existing UI against its own job and produce an evidence-backed, ranked findings report. Scopes the audit to named tasks and screens, runs a task walk, then checks visual hierarchy, layout and spacing, accessibility against WCAG 2.2 criteria, information density, cross-screen consistency, and conversion or workflow fit; separates defects from judgment calls; scores each finding by severity, reach, confidence, and effort. Use when someone asks for a design audit, UI review, UX heuristic review, an accessibility and consistency check of screens, or why an interface feels cluttered, confusing, or underperforms. Do not use for redesign directions (use explore-visual-directions), a funnel numbers diagnosis (use diagnose-conversion-funnel), a full site restructure (use map-site-for-redesign), or an accessibility test pass on built software (use accessibility-qa).
Audits a UI for hierarchy, layout, accessibility, density, consistency, and conversion or workflow fit.
Produce an information architecture and sitemap that people can find things in. Builds a content inventory with keep, merge, rewrite, or retire decisions; ranks top tasks; groups content by user mental models; chooses organization schemes; sets hierarchy depth and breadth; specifies global, local, and contextual navigation; writes labeling rules and final labels from user vocabulary; drafts URL paths and a redirect map; and plans card-sort and tree-test validation. Use when someone asks for a sitemap, information architecture, navigation structure, menu labels, how to group or reorganize site or app content, or why users cannot find things. Do not use for full page-by-page marketing site copy (use create-marketing-site-blueprint), the view inventory of an application from a PRD (use map-application-views), screen-level layout (use wireframe-spec), or capturing a live site for redesign (use map-site-for-redesign).
Produces information architecture, sitemap, navigation, labeling, and content grouping recommendations.
Write a prototype brief that ties a build to one decision. Covers the learning question and decision owner, interaction goals, scenarios with start and end states, a fidelity choice made per dimension (visual, interaction, content, data), a scope cut of what is real, faked, or narrated, an asset and data list, a state inventory, feedback prompts, evidence thresholds, and a build timebox. Use when someone asks for a prototype brief, what to prototype, how high fidelity a prototype should be, what a clickable prototype must cover, or what to ask people while they try it. Do not use for a paste-ready prompt for a design generator (use create-wireframe-prompts), a screen-by-screen wireframe specification (use wireframe-spec), the test protocol itself (use usability-test-plan), or visual style exploration (use explore-visual-directions).
Writes prototype briefs with interaction goals, scenarios, fidelity, assets, states, and feedback prompts.
Create a moderated or unmoderated usability test plan that a team can run and trust. Fixes the research questions and decisions, chooses the method, defines participant criteria and a screener, sets a sample-size rationale, writes realistic scenario-based tasks with success, assist, and failure definitions, selects metrics and a data-capture sheet, supplies the moderator script and probes, schedules a pilot, and pre-states the analysis rules, severity scale, and write-up format. Use when someone asks for a usability test plan, a user testing script, tasks and success criteria for testing a prototype or product, how many participants to test, or how to report usability findings. Do not use for planning broader discovery research (use ux-research-plan), writing the prototype under test (use prototype-brief), auditing screens without users (use design-audit), or analyzing survey exports (use analyze-survey-results).
Creates moderated or unmoderated usability test plans with tasks, success criteria, data capture, and reporting.
Create a UX research plan that connects a product decision to evidence. Turns decisions into answerable research questions, separates known facts, assumptions, and unknowns, writes falsifiable hypotheses with pre-set decision thresholds, selects methods by question type, defines participant criteria with a screener and a sample-size logic, drafts the interview or session guide, plans consent and data handling, builds a backwards timeline, and specifies the synthesis approach and readout. Use when someone asks for a UX research plan, a discovery or generative research study, interview plan, research hypotheses, which research method to use, or how to recruit and what to ask users. Do not use for a usability test protocol with tasks and success criteria (use usability-test-plan), analyzing an existing survey export (use analyze-survey-results), synthesizing feedback already collected (use synthesize-product-feedback), or controlled experiments (use ab-test-setup).
Creates UX research plans with goals, hypotheses, participant criteria, methods, script, and synthesis plan.
Create a reviewable visual-style exploration for an existing product, page, or interface before implementation. Use when someone asks to see how a current experience would look in a named aesthetic or described vibe, such as glassmorphism, editorial, brutalist, retro, premium, playful, or futuristic. Produces an isolated preview or implementation-ready visual specification without replacing the established brand system or changing production.
Create a reviewable visual-style exploration for an existing product, page, or interface before implementation. Use when someone asks to see how a current experience would look in a named aesthetic or described vibe, such as glassmorphism, editorial, brutalist, retro, premium, playful, or futuristic. Produces an isolated preview or implementation-ready visual specification without replacing the established brand system or changing production.
Create one paste-ready prompt for Claude Design or another design generator to produce every required low-fidelity wireframe from a PRD, functional requirements, and optional application-view inventory. Use for wireframe-generation handoffs, not when the user wants the wireframes rendered directly or high-fidelity visual design.
Create one paste-ready prompt for Claude Design or another design generator to produce every required low-fidelity wireframe from a PRD, functional requirements, and optional application-view inventory. Use for wireframe-generation handoffs, not when the user wants the wireframes rendered directly or high-fidelity visual design.
Turn product needs into wireframe-level screen specifications a designer and developer can both build from. For each screen it fixes the purpose and primary action, entry and exit points, content priority, layout regions in reading order, component contents and data sources, interaction rules as trigger-condition-response rows, a full state matrix, validation and permission rules, responsive reflow behavior, accessibility annotations, and requirement traceability. Use when someone asks for a wireframe spec, screen specification, annotated wireframes, interaction notes, states and responsive behavior for a screen, or a spec to hand to engineering. Do not use for a paste-ready prompt to a design generator (use create-wireframe-prompts), deriving the list of application views (use map-application-views), a prototype plan (use prototype-brief), or site-level navigation and labeling (use information-architecture-sitemap).
Turns product needs into wireframe-level screen specs, interaction notes, states, and responsive behavior.
Screenshot an entire existing web app and organize it by user flow into a design-handoff pack for redesign or UX optimization work. Discovers every route from source files or a link crawl, runs the app in an isolated environment with seeded fixture data and personas, captures every route per persona and desktop/mobile viewport, deep-captures tabs, modals, empty, loading, error and permission states and multi-step flows, writes a spec per page and end-to-end flow docs with screenshot references, and packages a master brief, per-area prompts, contact sheets and a findings list (bugs, dead links, orphan and legacy pages) for AI design tools such as Claude Design or Figma. Use when asked to document, audit or screenshot every page of a product before a redesign. Do not use for marketing screenshot packs, for deriving views from a PRD with no running app, or for producing the redesigned screens themselves.
Screenshot an entire existing web app and organize it by user flow into a design-handoff pack for redesign or UX optimization work. Discovers every route from source files or a link crawl, runs the app in an isolated environment with seeded fixture data and personas, captures every route per persona and desktop/mobile viewport, deep-captures tabs, modals, empty, loading, error and permission states and multi-step flows, writes a spec per page and end-to-end flow docs with screenshot references, and packages a master brief, per-area prompts, contact sheets and a findings list (bugs, dead links, orphan and legacy pages) for AI design tools such as Claude Design or Figma. Use when asked to document, audit or screenshot every page of a product before a redesign. Do not use for marketing screenshot packs, for deriving views from a PRD with no running app, or for producing the redesigned screens themselves.