Image

Add the swarm-build animation to a website, where a swarm of named, coloured cursors draws boxes and builds a screenshot region by region, or rebuilds a whole live page block by block, like a multiplayer design file. Use when the user wants a hero or product image to "build itself", wants multiplayer or Figma-style cursors on a page, or names swarm-build. Works with plain HTML and any framework (React, Next.js, Vue, Svelte, Astro).

Spawn a complete, deployable digital company (a "firm") from a template firm made of Docker images, Colab/Kaggle notebooks, harness configuration and a values file. The model does the firm-specific creative work (agent roles and schemas, harness plugin selection, research and comparison of free platforms and providers, generation of secret-creation guides), fills the template with the firm's hard-coded values, validates it, and deploys it to the platforms the operator chose. Use this skill whenever the user wants to create, spawn, configure, deploy, clone, update, migrate or roll back a firm or company, asks which free platform or provider to use for inference, storage, hosting or compute, wants help creating API tokens or secrets for a firm, or mentions firm template, firm repo, Firm Forge, DSH harness plugins, notebook-based workers, or deploying agents to Hugging Face, Colab, Kaggle or a local machine. Use it even if the user does not say "skill" or "spawner".

Direcao criativa e producao visual por IA. Usar automaticamente quando pedirem criar, analisar, reproduzir, restaurar, editar ou aprimorar imagens, personagens, artes 3D, anuncios, logos e assets de jogos; preservar referencias, pesquisar quando relevante e revisar qualidade.

Digitize a scanned book PDF into a readable, searchable, selectable-text PDF with images and source-aware typography. Use for OCR, retypesetting, or restoring printed books from scans, especially with Mac Apple Vision plus PaddleOCR PP-StructureV3 cross-checks. Guide nontechnical users through simple font, size, and paragraph choices while handling OCR, QA, and difficult page layouts yourself.

Design an original, brand-matched icon pack (Figma components or a zip of SVGs) instead of reaching for an open-source set like Lucide, Material or Heroicons. Runs a short guided interview, scans the brand (logo, typeface, screenshots) to derive stroke, corners and mood, writes a style brief for approval, finds icons missing from the product's screens, offers metaphors for abstract ideas, draws test icons first, then builds and validates the full set, with mirrored RTL variants and culture-safe symbols. Use it whenever someone wants custom icons, an icon set, library or system for their product or design system, icons that "match our brand", a replacement for a stock icon library, RTL (Persian, Arabic, Hebrew) icon variants, or new icons added to a pack made earlier from a saved style brief, even if they never say "icon pack". Not for finding existing icons, designing a logo, or one-off illustrations.

Make PDF documents from HTML or Markdown and check how they look, read existing PDFs as text or page images, and merge PDFs, with the pdfpecker Python package (no browser, about a millisecond per page). Use when the user wants a PDF, such as an invoice, report, letter, certificate, ticket, label, CV or a Markdown document as PDF, or wants to preview, read, count pages of or merge PDFs.

Pack layered data into a PNG. The image humans see is the rendered SVG surface; the SVG source and a markdown document ride inside the file — in iTXt chunks that survive image pipelines and in a greppable trailer after IEND. Use when creating, unpacking, editing, or repacking sharethought files — self-contained artifacts for human-LLM collaborative exchange where the human reads the picture and the LLM reads the depth.

Adds a first-visit intro to a static or plain-HTML site. A start screen shows the owner's name, title and location; on start, a scan traces the real page layout and labeled cursor agents draw boxes, type text and reveal images to build the actual page, which ends flat and normal. Use when asked to add an "agentic build" intro, start screen or build animation to a website.

Find, look at, judge and organize photos in an Apple Photos library (or any folder of images) on a Mac. Use when someone wants to find a photo, picture, image, video, selfie or screenshot ("find that picture of...", photos from a trip, party, date or place, "where was I on", "how many from 2019"), search by what's in the picture (a private on-device embedding index), find near-duplicates or "more like this", sort or judge photos at scale with contact sheets, look at people/faces, receipts or text in photos, shared albums, Hidden or Recently Deleted, download an original that lives only in iCloud, import files and set their dates, Hidden or Favorite, or when osxphotos hangs or Photos.sqlite comes up.

Draws or exactly copies any image with math. Methods are Bézier vectorization, Fourier series (epicycles), signed distance functions (SDF), implicit/parametric/polar equations, Tupper's formula and other methods. Gives images and animations for social media and video editors (Reels, TikTok, Shorts, YouTube, transparent .mov/.webm), plus Desmos equations or GLSL. Use this skill when the user asks to draw, copy, trace, illustrate or animate something "with math", "with equations", "with Fourier", "with rotating circles", "with SDF", "in Desmos", "with Tupper's formula", or asks for math art, generative art, a name/logo/signature in epicycles, an image changed into equations, or visual math content, also when no method name is given.

Build presentation decks in the Meadow brand design system (black canvas, acid-lime #D3FF31 plate, Geist/Syne type, 1920×1080 grid) as PowerPoint .pptx, and optionally as editable Figma frames, plus image-generation prompts for the deck's backgrounds and illustrations. Use this whenever someone at Meadow asks for slides, a deck, a pitch, a commercial offer (КП), a case study, a one-pager in slide form, a client presentation, or wants an existing deck re-laid out "in Meadow style" — in Russian or

Analyse spreadsheets, Word/PowerPoint documents, PDFs, images and text files strictly on the local machine, with mandatory masking of personal data (names, phone numbers, addresses, national ID and passport numbers, bank cards, e-mail, plates, account IDs) in every mode, and with zero upload to cloud documents, cloud drives, knowledge bases, online converters, OCR services or external search. Use whenever a user hands over local files for statistics, extraction, profiling or summarisation under

Buat file Markdown berisi kumpulan "command" prompt gambar siap salin untuk ChatGPT (logo, carousel IG, story IG/WA, OG image, feature image artikel, X/Twitter, LinkedIn, Pinterest, iklan, testimoni, mockup, dll) dari brand board, logo, foto, atau deskripsi brand yang diberikan user. Gunakan skill ini setiap kali user mengunggah brand identity board/logo/foto brand lalu meminta "buatkan command", "kumpulan prompt konten", "prompt sosmed", "command seperti /scrapbook tapi untuk brand saya", "temp

为用户指定的仓库生成配置驱动的浅色和深色 README SVG 横幅，选择 ASCII preset 与独立主题，按明确授权插入或更新 README 受管块。Use when a user asks to create, customize, regenerate, or check a README banner with readme-banner.

Stage 4 of a wp-site-kit WordPress build - real copy and photos. Content truth (facts only from the owner), AI drafts and owner approval, claims audits, banned words and fake proof, legal pages, the photo pipeline (dedupe, real-only, EXIF strip, WebP sizes, alt text, AI image register), and recorded content changes. Use for "write the copy", "add photos", "the owner sent new photos", "check for duplicates", "update prices or text on the live site", "legal pages".

Use when turning a CAD file (STEP/IGES) into product shots, transparent PNGs, or a 3D web embed.

Generates Open Graph / Twitter (X) share images for every page of a website and wires up the og:* and twitter:* meta tags, then verifies the social preview. Supports Next.js App Router (opengraph-image.tsx with next/og ImageResponse), Astro (build-time PNGs via satori + resvg) and any static site (Pillow script). Use when the user asks for OG images, social share cards, link previews, "my link looks bad on X/Slack/LinkedIn/WeChat", twitter:card, or before launching a site.

Writing image prompts — six building blocks in fixed order, word lists, negative prompts per model family, a length rule, and iteration with exactly one variable. Companion to the image-generation tooling.

THE LINEAGE, a LOWLIGHT STUDIO department (THE STUDIO). Reads a photograph on the user's machine (light, grain, framing, colour), finds the decade its look belongs to, traces its visual family tree back to 1930 through 107 named looks (films, cinematographers, photographers, processes), and grows the next branch: a 2027 look that no entry in the library has, with its recipe, a printable poster of the tree and a generation request for the user's own Higgsfield account. Use when the user says \"This shot didn't exist five minutes ago\", \"trace the lineage of this photo\", \"where does this look come from\", \"what decade is this photo\", \"grow the next branch\", \"the look nobody has made yet\", \"family tree of this shot\", or drops a still and asks for its references. Not for colour grading a whole video, identifying a specific film from a frame, screenshots of interfaces, illustrations or flat graphics (it reads light and film surface), or writing prompts to hand out.

Use when producing or updating App Store and Google Play screenshots and the Play feature graphic - planning a connected "panorama" story across frames, capturing real in-app screens with the app's own test/screenshot harness at each device's logical size and pixel ratio, covering the full device matrix (iPhone 6.9", the required medium Dynamic Island slot, iPhone Duo outer and inner displays, iPad 13" if the app runs on iPad, Play phone, 7" and 10" tablets, 1024x500 feature graphic), checking for alpha channels and wrong sizes, and uploading in order via the App Store Connect and Google Play APIs.

Turn one or more reference images into original art in that reference's own visual language, then cut it to whatever surface it's for. Use when you paste or point at art and say "use this as the base image", "generate art from this", "riff on this", "some creative ideation with these images", "make me a channel banner / avatar / cover / plate", "wilder version of this", or want a spread of options off one piece of reference. This is the TASK skill; the engine is the `make-art` tool skill.

Draw the way Leonardo da Vinci drew — a study sheet, a notebook folio, a red-chalk or silverpoint study, a machine cutaway, a storm of vortices — in his real materials and page-logic, generated with the make-art engine and anchored on fifteen of his actual (public-domain) sheets. Use when you say "Leonardo", "da Vinci", "like his notebooks", "a codex page of X", "draw this as one of his study sheets", "sanguine", "silverpoint", or wants a design, diagram, or illustration in that hand. Also a reading guide to his notebooks (Richter's 1883 public-domain translation) for "what did Leonardo say about shadow / trees / water". This is a TASK skill; the engine is `make-art`.

Turns a photo of a piece of furniture (a table, chair, bench, shelf, sideboard, bed, desk) into a buildable 3D model a carpenter can make from — every part as real stock with real joinery, an interactive offline HTML page with an exploded view, x-ray and step-by-step assembly, plus the maker pack - cut list, dimensioned shop drawings, what wood to buy, hardware and an illustrated assembly manual. Use when someone sends a furniture picture and wants it built, wants "plans", a "cut list", an "expl

Use when styling Zed like VS Code on Linux.