Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add OutlineDriven/odin-claude-plugin --skill frontend-ui-engineeringgit clone --depth 1 https://github.com/OutlineDriven/odin-claude-pluginWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/outlinedriven/odin-claude-plugin/frontend-ui-engineering)<a href="https://agentmods.dev/skills/outlinedriven/odin-claude-plugin/frontend-ui-engineering"><img src="https://agentmods.dev/badge/skills/outlinedriven/odin-claude-plugin/frontend-ui-engineering.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00046 | $0.01516 |
| Opus 5 | $0.00023 | $0.00758 |
| Sonnet 5 | $0.00009 | $0.00303 |
| Haiku 4.5 | $0.00005 | $0.00152 |
Grade A, and why
frontend-ui-engineering scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Frontend UI engineering
Contract
| Field | Bound contract |
|---|---|
| Trigger | Building or modifying user-facing interfaces and pages: creating components, implementing layouts, managing state, meeting WCAG requirements, or when the output must look production-quality and hand-crafted. |
| Authority | Reversible local: writes only named UI component and page files in the local working tree; rollback is undo. No remote mutation. |
| Side effect | UI components and pages created or modified in the local working tree. |
| Done | The UI meets the project design system, passes WCAG 2.1 AA keyboard and focus checks, is responsive at the required breakpoints, uses real content instead of placeholders, and rejects every named AI-aesthetic tell. |
Not for
- Token-system design with slop-audit gates: use frontend-design-deslop.
- Remote, credential, publish, deploy, or irreversible changes.
Inputs
Required:
- The component or page to build or modify, and the local file paths that will be created or changed.
- The project design system: spacing scale, semantic color tokens, type hierarchy, and border-radius scale. If absent, stop and request it; never invent values the system does not define.
- The required responsive breakpoints.
- Real content (actual labels, copy, and data shapes) for every rendered surface.
Optional:
- Framework and styling mechanism in use.
- State-management requirements and data-fetching approach.
Procedure
- Bound scope before writing: list every component and page file that will be created or modified. Do not touch files outside that list. Done when: the file list is stated and bounded.
- Read the project design system and use its tokens exclusively: spacing scale, semantic color tokens (
text-primary,bg-surface,border-default), type hierarchy, and border-radius scale. Never use raw hex values, off-scale pixel values, or invented spacing. Done when: every value traces to a design-system token. - Separate data fetching from presentation: a container component handles loading, error, and empty states; a presentational component receives resolved data and renders it. Done when: data fetching and presentation are in separate components.
- Compose small focused components rather than over-configured ones. Keep each component under roughly 200 lines and focused on one concern; split larger components by composition. Done when: no component exceeds 200 lines and each has one concern.
- Choose the narrowest state category that fits, in order: local component state, lifted to the parent, context, URL, server-state library, global store. Use local state for component UI, lifted state shared between two or three siblings, context for read-heavy write-rare values (theme, auth, locale), URL state for shareable filters and pagination, a server-state library for remote cached data, and a global store only for complex app-wide client state. Lift state or use context before prop-drilling past one level, and never prop-drill deeper than three levels. Done when: the state approach is the narrowest that fits and prop-drilling stays within its limits.
- Make every interactive element keyboard accessible. Use the native HTML element first (
<button>,<a>,<input>) so elements are focusable by default; reach for ARIA only when no native element fits. If a non-interactive element must act as a control, addrole,tabIndex, andEnter/Spacekey handlers. Providearia-labelfor icon-only controls and for inputs with no visible label; pair inputs with<label htmlFor>. Done when: every interactive element is keyboard accessible with a visible label or aria-label. - Manage focus when content changes: move focus to newly revealed content or its close control, and trap focus inside modal dialogs while they are open. Done when: focus management is implemented for every dynamic content change and modal.
- Render meaningful empty, loading, and error states for every data-driven surface. Use skeleton placeholders marked
aria-busy="true"for loading, never blank screens or spinners for content areas. Never use color as the sole indicator of state; pair color with text or icons. Done when: all three states are rendered for every data-driven surface and no state relies on color alone. - Build mobile-first responsive layouts using the project breakpoint system, then expand upward. Verify the layout at 320px, 768px, 1024px, and 1440px, or at the project's required breakpoints if they differ. Done when: the layout is verified at all required breakpoints.
- Use real content everywhere. Placeholder or lorem-ipsum text hides wrapping, overflow, and length problems that real content reveals. Done when: every rendered surface uses real content.
- Reject the AI aesthetic with concrete tells: no default purple or indigo palettes, no excessive gradients, no
rounded-2xlon everything, no oversized uniform padding, no stock card grids, no generic hero sections where content-first layouts serve, and no layered shadows unless the design system specifies them. Use the project's actual palette, flat or subtle gradients matching the system, and consistent border-radius from the system. Done when: every AI-aesthetic tell is checked and rejected. - Verify before declaring done: render without console errors; Tab through every interactive element; confirm a screen reader can convey structure; confirm responsiveness at the required breakpoints; confirm loading, error, and empty states are handled; confirm design-system adherence; run axe-core or browser dev-tools accessibility scan and resolve every warning. Done when: every check passes with zero warnings.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 62 lines · 46 tokens per session scan A 202881659983
frontend-ui-engineering is a skill published in the GitHub repository OutlineDriven/odin-claude-plugin (35 stars, last pushed today), licensed Apache-2.0. It adds 46 tokens to every session and 1,516 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-06.
Other skills, from other repositories
aceternity-ui
100+ animated React components (Aceternity UI) for Next.js with Tailwind. Use for hero sections, parallax, 3D effects, or encountering animation, shadcn CLI integration errors.
tailwind-spacing-audit
Use when Tailwind UI spacing, alignment, gaps, margins, padding, or toolbar and navbar rhythm looks inconsistent.
ui-polish
Design-engineering details that make an interface feel polished — border radius, optical alignment, shadows and elevation, animations and micro-interactions, press feedback, icons. INVOKE PROACTIVELY when building or reviewing UI components, adding motion or hover/active states, or when the user says "make it feel…
layout
Layout structure for web interfaces — grouping, alignment, negative space, reading order, progressive disclosure, breakpoints, direction-aware (RTL) structure. INVOKE PROACTIVELY when structuring a page or component, spacing or aligning controls, deciding what collapses at small sizes, or reviewing frontend structure.…
svg-mastery
Expert knowledge for working with SVG (Scalable Vector Graphics) — optimization, embedding, animation, accessibility, responsive scaling, React integration, filters, and programmatic creation. Use when the user asks to "optimize SVG", "clean up SVG", "reduce SVG file size", "embed SVG in HTML", "inline SVG vs img…
frontend-aesthetics
Design guidance and aesthetic patterns for building beautiful, distinctive frontends — websites, web apps, landing pages. Provides typography strategies, color application, motion choreography, layout composition, and component design recipes. Use when the user asks to "make this look better", "improve the design"…