magiccube/agentara/glassify-ui
Claude Code and agent skill by magiccube/agentara
- community
Checking that the SKILL.md file is still in the repository.
Transform a UI screenshot into a structured Cinema4D/Octane JSON render prompt for photorealistic 3D poster art. Use this skill whenever the user uploads a UI screenshot (app, website, tool, dashboard) and wants a Cinema4D render prompt, 3D poster, AIGC image prompt, or any photorealistic visualization of an interface. The output is always a clean JSON file — never prose or Midjourney text strings.
Install
npx vskill@latest install magiccube/agentara/glassify-uiSKILL.md
Cinema4D UI Render Skill
Transform a UI screenshot into a structured JSON render prompt for Cinema4D/Octane. Output purpose: promotional poster art, not documentation. Abstraction beats fidelity.
Core Philosophy: "舍得" (Less is More)
A poster is not a screenshot. The JSON you produce must:
- Simplify layers — 4–8 layers max. Merge minor UI regions. Omit decorative chrome.
- Abstract content — No literal button labels unless they define the product identity.
Use semantic descriptions: "primary action buttons row" not "Bold / Italic / Underline".
- Elevate the hero — One layer is the visual anchor (the "hero slab"). Give it the
prominent glow, active color, and bloom effect. Everything else supports it.
- No clutter — Status bars, scrollbars, tooltips, and minor icons: omit entirely
unless they are distinctive brand elements.
- Extract interactive controls as micro_elements — Every visible button, toggle switch,
pill chip, or CTA must be listed in the micro_elements array as an independent glass brick with its own z_offset, shape, and tint. Do NOT bury them inside a parent layer's content description. This is mandatory — missing micro_elements is a critical error.
Step 1 — Analyze the Screenshot
Look at the screenshot and extract:
- UI category: productivity / developer / creative / mobile / web / dashboard
- Brand identity signals: dominant color, logo, product name, characteristic icon
- Natural focal region: which area draws the eye first — that becomes the hero layer
- Aspect ratio decision (if user has not specified):
- Desktop/web UI with wide layout →
"16:9" - Mobile app or portrait-first UI →
"9:16" - Square or ambiguous →
"1:1"or"16:9"as default
Step 2 — Design the Layer Stack
Collapse the real UI into 4–8 abstract depth layers, Z-offset spaced 5–6mm apart. Use this table as a starting template — adapt to the actual UI:
| Layer ID | Typical Role | Z offset | Thickness | Roughness | Transmission | |
|---|---|---|---|---|---|---|
| 0 | App shell / chrome | 0mm | 8mm | 0.20 | 0.50 | |
| 1 | Background / canvas | 5mm | 6mm | 0.15 | 0.65 | |
| 2 | Main content area | 10mm | 5mm | 0.06 | 0.85 | |
| 3 | Primary toolbar/nav | 15mm | 5mm | 0.12 | 0.70 | |
| 4 | Hero element | 20mm | 4mm | 0.08 | 0.75 | |
| 5 | Floating accent (opt.) | 25mm | 3mm | 0.18 | 0.60 |
Rules:
- The hero layer gets
glowandactive_elementwitheffect: "backlit_halo_bloom" - Content descriptions use semantic phrases, never exhaustive lists of UI elements
- Omit Layer 5 if the UI has no natural floating element
micro_elements (mandatory if interactive controls exist)
After the main layer stack, scan the screenshot for every interactive widget: buttons, toggle switches, pill chips, dropdown triggers, submit CTAs, radio buttons, sliders. Each one becomes an entry in micro_elements — a separate floating glass brick:
type values: button_primary, button_secondary, toggle_switch, pill_chip, dropdown_trigger, icon_button, cta_submit. shape values: rectangle, pill, circle, square. tint values: accent_blue, accent_teal, neutral_white, clear, warm_white, cool_white.
Step 3 — Choose Environment & Tone
Default white balance is neutral-to-slightly-cool (5200–5600K). Avoid warm golden tones unless explicitly requested. Match environment to UI category:
| UI Category | Environment | Tone | Palette | |
|---|---|---|---|---|
| Productivity | bright_minimal_desk_daylight | neutral_cool | clean_daylight | |
| Developer / Code | dark_minimalist_desk_led_strip | cool | cool_white_led | |
| Creative / Design | architects_drafting_table | neutral_cool | soft_overcast | |
| Mobile App | white_marble_surface_lifestyle | neutral_cool | clean_daylight | |
| Dashboard / Data | glass_conference_table | cool | morning_cool_light |
Props are 2–3 contextually appropriate objects, soft bokeh background. UI float height: 15–20cm above surface.
Step 4 — Output the JSON
Produce exactly this structure. No extra fields, no prose, no markdown around the JSON.
Quality Checklist (self-review before output)
Before writing the final JSON, verify:
- [ ] Layer count is 4–8, not more
- [ ] Each
contentarray has ≤ 3 items, all semantic (no literal button labels) - [ ] Exactly one layer has a
glowvalue and anactive_element - [ ]
hero_layer_idmatches that layer'sid - [ ]
focus_targetin camera matcheshero_layer_id - [ ] Environment and tone are consistent — default is neutral_cool, never warm/golden unless user asked
- [ ]
white_balance_kelvinis set (5200–5600K range for neutral cool) - [ ]
negative_promptarray is present and includes annotation-text suppressors - [ ]
micro_elementsarray is present if the UI has any buttons, toggles, or chips - [ ] Every button / toggle / pill in the screenshot has a corresponding
micro_elemententry - [ ]
style_tagsincludes"light_mode_ui"and"no_annotation_text" - [ ] Aspect ratio is decided and justified
- [ ] No
midjourney_suffixor prose fields present
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckinguser-home/.claude/skills/glassify-ui/SKILL.md
Popularity
- Repository stars
- 515
- Repository forks
- 58
- 7-day trend score
- 31
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 26 Sep 2026, a month ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/magiccube/agentara/glassify-ui)