AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Rlabs Design

skill-0xrafasec-ai-workflow-rlabs-design · by 0xrafasec

Use this skill to generate well-branded interfaces and assets for Rlabs, either for production or throwaway prototypes/mocks/etc. Contains essential design guidelines, colors, type, fonts, assets, and UI kit components for prototyping.

No reviews yet
0 installs
31 views
0.0% view→install

Install

$ agentstack add skill-0xrafasec-ai-workflow-rlabs-design

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-0xrafasec-ai-workflow-rlabs-design)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Rlabs Design? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Read README.md first — it carries the brand voice, palette rules, motion grammar, accessibility floor, and caveats. Then load the materials that match the surface you're building. Skipping the matching kit is the most common source of off-brand output.

Foundations (load on every surface)

  • colors_and_type.css — light + dark CSS variables and semantic tokens.
  • tokens.css — spacing, radii, elevation, motion tokens.
  • assets/logo/rlabs-logo.svg, rlabs-mark.svg. Use these directly; never recreate.
  • assets/brand/chain-glyph.svg — the chain-of-nodes motif. Use for hero/section accents only.
  • assets/icons.md — Lucide setup (24px, 1.75px stroke). Stroke-only.

Surface-specific materials (mandatory before designing)

Pick the bucket that matches what the user asked for and load everything in it. Reading just the CSS tokens is not enough — the kits encode component composition, spacing rhythm, and copy patterns that the tokens alone don't convey.

  • Slides / presentations / decks → load slides/index.html end-to-end. It is the canonical 7-slide template (title, section, content, metric, quote, comparison, closing) — copy structure and class names from it instead of inventing new ones.
  • Marketing website / landing pages → load ui_kits/website/index.html, then read each component: SiteNav.jsx, Hero.jsx, FeatureGrid.jsx, CodeShowcase.jsx, Footer.jsx, plus README.md.
  • Product app / dashboard / console → load ui_kits/app/index.html, then read Sidebar.jsx, TopBar.jsx, Overview.jsx, Agents.jsx, Transactions.jsx, _shared/Primitives.jsx, plus README.md.

Preview specimens (consult before inventing)

Before composing a component, check preview/ for the matching specimen card and match its proportions, padding, and state treatment. These are the source of truth for token application:

  • Color: colors-core.html, colors-neutrals*.html, colors-accents.html, colors-semantic.html, colors-modes.html.
  • Type: type-display.html, type-body.html, type-mono.html, type-scale.html.
  • Spacing & shape: spacing-scale.html, spacing-radii.html, spacing-elevation.html, spacing-motion.html, radii.html, elevation.html.
  • Components: components-buttons.html, components-inputs.html, components-nav.html, components-card.html, components-cards.html, components-badges.html.
  • Brand application: brand-logo.html, brand-logo-dark.html, brand-logo-light.html, brand-glyph.html, brand-motif.html, brand-icons.html, brand-voice.html.

If a token decision feels ambiguous, the preview card wins.

Output mode

  • Visual artifact (slides, mocks, throwaway prototypes): copy assets out and emit static HTML the user can open. Reuse the template/kit markup verbatim — adjust content, not structure.
  • Production code: copy the CSS variables and component patterns into the target codebase. Don't re-derive tokens; reference the existing files.

If invoked with no guidance, ask the user which surface (slides / website / app) and what they want to build, then load the matching kit before producing anything.

Mandatory verification — Playwright fidelity check

After applying the design (artifact or production code), you must run a Playwright check before reporting done. Visual drift between intent and output is the failure mode this skill exists to prevent.

Use the playwright MCP (mcp__plugin_playwright_playwright__*) — it is available in this environment.

  1. Open the reference in a tab: the matching index.html from slides/, ui_kits/website/, or ui_kits/app/. Take a full-page screenshot.
  2. Open your output in a second tab. Take a screenshot at the same viewport (default 1440×900 desktop; also 390×844 if mobile is in scope).
  3. Compare against the brand rules — not just pixel-diff. Check:
  • Typography: Poppins ExtraBold display, Roboto body, JetBrains Mono for hashes/code. Lowercase OR uppercase headings (never title case).
  • Palette: black/white first; ≤1 accent per view; correct surface (#0A0A0B dark, #FAFAF7 light).
  • Radii: 0/4/8/12 only. No pills outside status chips.
  • Spacing: 4px grid (4, 8, 12, 16, 24, 32, 48, 64, 96, 128).
  • Borders: hairline rgba(_,0.08) defaults.
  • Motion: hover/press present and on cubic-bezier(.2,.8,.2,1); no bounces.
  • Copy: no emoji, no title case, no hype words, for CTAs.
  1. Fix any drift in place, re-screenshot, and only then report done. Note the screenshots taken in your final message.

If Playwright is unavailable in the current environment, say so explicitly and list the brand-rule checks you performed by reading the output instead — do not silently skip verification.

Quick cheat sheet

  • Headings: Poppins ExtraBold 800, lowercase (or UPPERCASE), tight tracking (-0.02em).
  • Body: Roboto, 400 default. Medium (500) for UI labels. Light (300) for ledes.
  • Mono: JetBrains Mono for hashes, code, tabular numbers.
  • Black/white first. Accents: Signal #C6F24E, Ember #FF5B2E, Lapis #2E5BFF, Violet #8B5CF6. One accent per view.
  • Sharp corners (radius 0–12px; never >12). No rounded pills except status chips.
  • Motion: cubic-bezier(.2,.8,.2,1), 120/200/320/520ms. Transform + opacity only. No bounces. No linear easings.
  • No emoji in product or marketing copy. Unicode arrows () encouraged.
  • Dark surface: #0A0A0B (not pure black). Light surface: #FAFAF7 (warm paper).

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.