assets
Skill1 starsChanged 53 days agoMITView on GitHub
Produce a web-kit site's visuals. Covers the user's own photos and video (selecting, compressing, stripping GPS data, embedding), CSS/SVG motion, images and animations generated with Higgsfield from DESIGN.md's art direction, and the social preview image. Use it for any hero, gallery, portfolio, reel, illustration, background, video, or animation, when a spec slot says "asset:", when a section looks empty or generic, when the user has photos or footage to show, or when they mention Higgsfield or the image shown when a link is shared.
- Installs packages
10 other projects use a version of this.
Adopted by
Most recently taken up by
- PicsArt/gen-ai-skillstheir own version, first seen Nov 8, 2026
- rylsherdamz-rgb/stellar-forgetheir own version, first seen Nov 8, 2026
- BuilderIO/agent-nativetheir own version, first seen Oct 8, 2026
- ultimatethemedev-bit/claude-vitrine-kittheir own version, first seen Sep 20, 2026
- SamRome1/showrunnertheir own version, first seen Sep 14, 2026
- pedrovelasquez9/fraguatheir own version, first seen Aug 15, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- BuilderIO/agent-native★ 7.6ksame name
- Kucoin/kucoin-skills-hub★ 28same name
- spotify/ads-agentic-tools★ 20same name
- rylsherdamz-rgb/stellar-forge★ 17same name
- pedrovelasquez9/fragua★ 12same name
- PicsArt/gen-ai-skills★ 6same name
- alanvaa06/award-craft★ 1same name
- emretasss/stellar-paygate-mcp★ 1same name
- SamRome1/showrunner★ 1same name
- ultimatethemedev-bit/claude-vitrine-kit★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository7 other files it gives its agents.
- template/CLAUDE.mdCLAUDE.md
- design-loopSkill
- design-systemSkill
- new-siteSkill
- publishSkill
- retroSkill
- site-copySkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 3 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 8 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
CSS/SVG: Use it for micro-motion, hover, reveals, diagrams, and patterns.
- Distinctive: line 14
Generated with Higgsfield: Use it only with a paid budget (README.md §9), for what the user can't shoot: a hero concept, illustrations, a …
- Distinctive: line 28
Keep a hero under 300 KB and a card under 120 KB.
- Distinctive: line 57
Keep the subject centered, because some apps crop it to a square.
- Very specific: line 61
Use <img> with width and height (no layout shift).
“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Assets
Generic icons and stock visuals are the fastest tell of an AI-made site. The most convincing visual is usually the user's real work. Everything else comes from one art direction, so the set looks coherent.
Choose the source (best first)
- The user's own photos and video of their real work, team, place, or product. Ask for them early, and say exactly what works: wide shots with room for text, good light, no watermarks, and landscape for heroes. They're free and they build trust that no generated image can.
- CSS/SVG. Use it for micro-motion, hover, reveals, diagrams, and patterns. It's free, sharp, and themeable with tokens. Follow the Motion section of
DESIGN.md. - Generated with Higgsfield. Use it only with a paid budget (
README.md§9), for what the user can't shoot: a hero concept, illustrations, a mascot, textures. Before generating anything, list the assets and roughly how many credits they'll use, and wait for a yes.
With a free budget, use options 1 and 2 only. That's a complete, professional result, not a lesser version.
Tools
The user's media needs ffmpeg, installed once: Windows winget install Gyan.FFmpeg, macOS brew install ffmpeg, Linux apt install ffmpeg. The originals go in media-originals/, which is gitignored because it's too heavy for git. The web versions go in public/assets/<page>-<slot>.<ext>.
Own photos
- Select 1 strong image per slot rather than many average ones. For a gallery or portfolio, pick 6–12 of the best, grouped by service.
- Compress and strip metadata. Drone and phone photos carry GPS coordinates in their EXIF data, so strip it before publishing:
ffmpeg -i media-originals/x.jpg -map_metadata -1 -vf "scale='min(2400,iw)':-2" -c:v libwebp -quality 80 public/assets/x.webpAlso make a 1200 px version for cards and mobile. Serve both withsrcset. - Keep a hero under 300 KB and a card under 120 KB.
Own video
- Hero or background loop: 6–15 s of the best footage, silent, 1080p at most:
ffmpeg -i media-originals/x.mp4 -map_metadata -1 -an -vf "scale=-2:1080" -c:v libx264 -crf 26 -preset slow -movflags +faststart public/assets/x.mp4Make a poster frame too:ffmpeg -ss 1 -i public/assets/x.mp4 -frames:v 1 -c:v libwebp public/assets/x-poster.webp. Aim for under 5 MB. Use<video autoplay muted loop playsinline preload="metadata" poster=…>. Underprefers-reduced-motion, show only the poster. - Long reels, showreels, or anything with sound: upload to YouTube or Vimeo (free). Embed as a poster image with a play button. Load the
<iframe>only on click (youtube-nocookie.com), so the page stays fast and third-party cookies don't load until the visitor asks for the video. - Hard limit: Cloudflare Pages rejects any file over 25 MiB. Anything larger goes to YouTube or Vimeo.
Generated images and video (paid budget)
Setup, once per machine: npm i -g @higgsfield/cli, higgsfield auth login, npx skills add higgsfield-ai/skills. Higgsfield's own skills teach the exact commands and models. Without the CLI, write the prompts to docs/assets.md for the user to run in the Higgsfield web app.
Build every prompt from DESIGN.md → Art direction, in this order:
<subject + action> · <composition: framing, negative space for copy on the left or right> · <style: abstraction level, material, rendering> · <lighting> · <palette: 3–5 hex from the tokens> · <mood> · avoid: <forbidden clichés from DESIGN.md>, text, logos, watermarks
- Reuse the same style, lighting, and palette block word for word across all prompts. Only the subject and composition change.
- Match the aspect ratio to the slot: 16:9 or 21:9 for a hero, 4:5 for cards, 1:1 for marks.
- Generate 3–4 variants. Pick the one that best fits the section in the running page. Compress it the same way as own media.
- Video only where motion carries the message: 5–10 s, loopable, silent.
Social preview image (every site)
public/og.jpg at 1200×630 is what WhatsApp, LinkedIn, and others show when a link is shared. Use the hero photo, or a strong own photo, cropped:
ffmpeg -i media-originals/x.jpg -map_metadata -1 -vf "scale=1200:630:force_original_aspect_ratio=increase,crop=1200:630" -q:v 3 public/og.jpg
Keep the subject centered, because some apps crop it to a square. The launch check reports it if it's missing.
Markup and log
- Use
<img>withwidthandheight(no layout shift). Take thealttext fromsite.jsin every language, and give decorative imagesalt="". Addloading="lazy"below the fold. - Append a row to
docs/assets.md(create it with this header if missing):| File | Slot | Source (own / CSS / model) | Original or prompt | Date |. That way any asset can be traced, regenerated, or replaced later.
Then run web-kit:design-loop on the section, so the asset is judged in place.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

