用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/SocketDev/action --skill designing-interfaces命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Propagate a wheelhouse template change across fleet repos: worktrees, push/PR fallback, cleanup.
Run this repo's GitHub Actions locally with Agent-CI before pushing CI-sensitive changes.
Audit package exports for dead, internal-only, or weakly-consumed subpaths before pruning.
基于 SOC 职业分类
正在显示 SKILL.md
| name | designing-interfaces |
| description | Designs and reviews accessible interfaces using reference research and rendered validation. |
| license | MIT |
| compatibility | Useful on its own as a design methodology and craft reference; works best with Refero MCP available for live style, screen, and flow research. |
| metadata | {"internal":true,"author":"SocketDev","sources":"Refero skill @ f78b4ecc (2026-07-10)"} |
This is the fleet's consolidated interface-design workflow. Refero supplies the reference-research capability; the companion skills own focused implementation, review, performance, and test work.
Refero gives agents taste and product evidence. Use it before design work instead of relying on generic model knowledge.
Refero has three research layers:
Best results come from combining layers: visual direction from styles, concrete UI patterns from screens, and sequencing from flows when the task has multiple steps.
This skill owns design direction: reference research, the visual lock, and the implementation direction. Use the companion skills only for their narrower roles:
This skill is useful on its own as a research-first design methodology and craft reference. Research is mandatory. Use Refero MCP for live style, screen, and flow research when available; otherwise research with bundled craft references and any user-provided references.
Typical MCP setup:
claude mcp add --transport http refero https://api.refero.design/mcp --header "Authorization: Bearer <token>"
For full tool details, read references/mcp-tools.md.
Before researching, form a short design brief. Ask only for missing information that would materially change the result; otherwise make reasonable assumptions and proceed.
Clarify:
Brief format:
Designing [WHAT] for [WHO] on [PLATFORM].
Goal: [PRIMARY USER GOAL].
Tone: [DESIRED FEELING].
Main objection/risk: [OBJECTION].
Must remember: [HOOK OR DISTINCTIVE IDEA].
Constraints: [CONSTRAINTS].
Research needed: [styles/screens/flows].
Path: [direct build / visual exploration / audit / asset generation].
Choose the lightest workflow that can produce a high-quality result.
Use refero_search_styles when the user asks to design, redesign, improve, polish, or
create anything with a visual component.
A style is a semantic design reference extracted from a real web marketing/product page.
It is not a screenshot and not a component library. Search results give previews; full
style references from refero_get_style provide design guidance such as visual thesis,
tokens, typography, layout/composition, section rhythm, spacing, elevation, surfaces,
components, imagery treatment, implementation notes, and do/don't rules.
Current limitation: Refero styles currently cover web marketing/product pages such as landing pages, pricing pages, product marketing sites, editorial brand sites, and SaaS websites. They do not currently cover in-app dashboards, auth screens, settings screens, or iOS app screens as style systems. Still use styles for product UI tasks to establish visual language, then use screens/flows for product logic.
Use styles for:
Use refero_search_screens when you need:
After finding strong screens:
refero_get_screen for full detailsrefero_get_similar_screens to expand from a strong examplerefero_get_screen_image only when raw screenshot inspection is neededUse refero_search_flows when the task has a before/after sequence:
After finding a strong flow, use refero_get_flow for step-by-step goals, actions,
system responses, and completion states.
For image generation, visual options, generated assets, and visual QA, read references/visual-workflow.md when the task needs it.
For any visual design task, start here.
Recommended loop:
refero_get_style; full styles are large, so split larger research into multiple batches.Good style queries:
Extract from styles:
Synthesis rule:
Never present the result as "copying X". Present it as a new direction inspired by several references.
Before implementation, create a reference lock:
Primary reference/direction: [one dominant source]
Preserve: [3-5 traits that must survive: canvas, type, accent, layout, density, media]
Borrow only: [1-2 specific secondary details]
Role rules: [source token/component meanings to preserve, e.g. CTA-only, code-only, decorative-only]
Media strategy: [real/generated/stock/code-native/placeholder, with aspect ratio and art direction]
Reject: [defaults/averages that would collapse the direction]
Token commitments: [background, type, accent, radius, border/shadow, imagery treatment, with roles]
If implementation drifts from the lock, stop and correct it. Do not soften distinctive traits into safer colors, safer fonts, softer radius, or generic section layouts. Reference lock is not cloning; it preserves selected traits while adapting content, brand, and interaction details to the user's product.
When combining styles, assign each source a bounded job. For example: one source may own canvas/type, another may own code-window treatment, and another may own primary CTA. Never move a token outside its source role: CTA colors stay CTA-only, syntax colors stay inside code, decorative gradients stay decorative, and card/button rules keep their specified radius, shadow, and state behavior.
If the primary style is image-led, do not replace it with text-only layout. If you cannot produce the needed image or graphic, preserve the slot with stable dimensions, aspect ratio, caption/alt text, and a short art-direction note. Build simple diagrams, icons, code windows, or geometric primitives only when they match the source style.
For substantial visual exploration, generated mockups, bitmap assets, or post-build visual QA, follow references/visual-workflow.md.
Use screens when you need to know what the interface should contain or how real products solve a specific UI problem.
Good screen queries:
Search by facts on the screen:
Avoid using screens as the primary style source when the task is visual. Use styles first, then screens for structure and concrete details.
Extract from screens:
Use flows when there are multiple steps or a user changes state over time.
Good flow queries:
If flow search is sparse, broaden the query. If still sparse, use screens and reconstruct the journey.
Extract from flows:
Match depth to task risk.
For a quick visual improvement:
For a new landing page, brand direction, or major redesign:
For a product workflow:
For high-stakes or ambiguous tasks:
Separate findings into three buckets.
From styles:
Output example:
Use a precise analytics SaaS foundation: white canvas, compact UI copy, restrained black
primary actions, thin borders, and product screenshots in framed panels. Borrow disciplined
accent use from another reference, but keep color rare.
From screens:
Output example:
Pricing pages commonly put the billing toggle above plan cards, highlight one plan, and
move detailed feature comparison below. We should adapt the comparison structure but keep
the hero quieter because this product sells trust, not hype.
From flows:
Output example:
Cancellation flows usually collect a reason, offer a relevant alternative, confirm the
destructive action, then state when access ends. The best flows give a clear return path.
Before presenting research, implementing the direction, or doing final visual QA, read delivery-and-craft.md.
For UI labels, errors, empty states, confirmations, and recovery copy, read interface-copy.md.
For final tuning dials like clarify, distill, harden, onboard, bolder, quieter, and delight, read refinement-modes.md.