design-and-ai-skills

Use when the user invokes /design, or asks to design/build/audit/polish a web or cross-platform UI or create a design case study — screen, component, layout, table, chart, form, color system, portfolio narrative — and it's unclear which design reference skills apply. Dispatcher that routes to the right companion skills, delegates case studies to /case-study, and routes native iOS work to /ios.

Use when the user invokes /ios, or when starting ANY native iOS/SwiftUI screen, component, flow, or review and it's unclear which ios-* skills apply. Dispatcher that routes to the right 2-4 companion ios-* reference skills instead of loading the whole family or designing from memory.

iOS semantic system colors, dark mode, and material levels per Apple HIG (current iOS 26). Covers why you never hard-code system color values, the system vs grouped background sets, label/separator colors, dark mode base-vs-elevated + the 4.5:1/7:1 contrast floors, the four standard material levels + vibrancy, Liquid Glass color rules, and Display P3. Use when choosing colors, backgrounds, or materials for any native iOS screen.

Apple HIG component rules for native iOS 26 — bars (tab/nav/toolbar/search incl. floating + concentric corners), buttons (44pt hit region, prominence, roles, no destructive-primary), menus/pull-down/pop-up, toggles/pickers/segmented/sliders/steppers/text fields, lists & tables, sheets + detents, alerts vs action sheets vs popovers, modality, progress/gauges, with the SwiftUI mapping. Use when choosing or reviewing any iOS control or presentation.

Apple HIG behavioral patterns for native iOS — onboarding, launch screens, loading, modality decisions, feedback, settings, search, privacy & permission flows (purpose strings, priming screens, ATT), account deletion, in-app purchase, ratings, notifications — plus motion/reduce-motion, haptics (UIFeedbackGenerator patterns), and the accessibility floors (contrast AA, touch targets, Dynamic Type 200%, VoiceOver). Use when designing any flow, permission prompt, empty/loading state, or feedback moment on iOS.

iOS layout & spacing for iPhone — the 8pt grid convention + 4pt sub-grid, real Apple layout numbers (16/20pt system margins, safe-area insets, 44/49/50pt bar heights), the 44pt touch target, building dense low-scroll layouts with LazyVGrid/Grid + GridItem (2-up/3-up rules), SwiftUI's adaptive default spacing, and the iOS 26 floating-chrome layout model. Use when laying out any iPhone screen, choosing margins/gutters, or building a grid.

iOS 26 Liquid Glass design language + the SwiftUI APIs that implement it, sourced from Apple's WWDC25 sessions 219/323 and developer.apple.com. Covers the functional-layer-over-content mental model, regular vs clear variants, glass-on-glass/over-tint don'ts, the floating tab bar, scroll edge effects, concentric corners, accessibility behavior, and a verified-vs-community API table. Use when building or reviewing any iOS 26 native screen's chrome, materials, or glass effects.

SF Symbols usage (weights, scales, the four rendering modes, fill-vs-outline variants, animations, SF Symbols 7 gradients/Draw, accessibility labels, version-gating) and iOS 26 app icons (layered design, Icon Composer, square-unmasked layers, the six appearances, specs) per Apple HIG. Use when picking icons, choosing rendering modes, or producing an app icon for a native iOS app.

iOS typography applied — the SF family system (SF Pro / Pro Rounded / Mono / New York), Dynamic Type built-in text styles + their point sizes, weights to prefer/avoid, type pairing and label-vs-value hierarchy for a minimal health/nutrition app, tabular/monospaced digits for data, and the concrete SwiftUI font APIs. Use when setting any type on a native iOS screen or building a type scale.

iPhone-specific Apple Human Interface Guidelines for iOS 26 (Liquid Glass) — the floating tab bar and how content scrolls under it, navigation/sheets/presentation detents, SF Symbols + Dynamic Type + safe areas, control styles, 44pt touch targets, and the concrete SwiftUI modifiers that implement each. Use when building or reviewing any native iPhone screen; pairs with platform-conventions (cross-platform) and components-and-states.

Designing mobile screens that fit one viewport with minimal vertical scrolling — viewport budgeting per device, above-the-fold prioritization, swapping tall stacks for grids/segmented views/disclosure/detented sheets, compact metric components, and when scrolling IS the right call. Use when a screen scrolls more than it should, or when condensing a feature-heavy phone screen; pairs with layout-and-composition, grid-and-spacing, ios26-hig-patterns.

UI/UX patterns for modern macro + food-tracking apps (MyFitnessPal, Cal AI, Lose It, MacroFactor, Lifesum) — the calorie/macro dashboard, the log→remaining loop, fast multi-modal entry (search/barcode/photo/voice), the food diary, weight trend, daily score, and the anti-patterns that make trackers feel like data entry chores. Use when designing or reviewing a nutrition/calorie app's screens; pairs with dense-no-scroll-layout, ios26-hig-patterns, data-visualization.

Reference-grade guide for AI engineers building autonomous and tool-using agents — the plan→act→observe loop and its failure physics, the five budgets every agent must set (iteration, tool-call, token, wall-clock, cost), termination and no-progress detection, runaway prevention (circuit breakers, kill switch), the three guardrail layers (input/output/action), verify-before-acting, bounded sub-agent orchestration, reflection cost, determinism-where-possible, checkpoint/resume, and the canonical failure modes with concrete budgets and pseudocode.

Reference-grade guide to LLM safety engineering and multi-tenant isolation for AI engineers and security teams — the untrusted-input threat model (OWASP LLM Top 10), direct and indirect prompt injection and why prompting alone can't solve it, defense in depth (privilege separation, dual-LLM, constrained action space, human approval), data-leakage and exfiltration prevention, permission boundaries where the system authorizes (not the model), and cross-tenant isolation across cache, vector index, and memory.