Creative OS

Tunes existing type to premium optical standards – tracking and line-height computed from font size instead of set as constants, the two-cut text/display split, the 600-weight ceiling, and the tuned-face spend ladder.

Disciplines layout space – the two-tier spacing system (4/8px micro-scale capped at 64px, then section cliffs at 96px and above), reading measure of 45-75 characters, container and target minimums.

Enforces premium color restraint on an existing palette at the client charter's recorded values – a default budget of one chromatic accent at 10 percent coverage or less with an operational coverage-measurement procedure.

Governs surfaces, depth, and effect scarcity – glass on sticky chrome and overlays only (one layer maximum, with a contrast fallback), dark-mode elevation via a 4-step surface ladder and 1px hairlines instead of shadows, whisper-quiet light-mode shadows.

Sets the physics that make motion feel premium – the productive/expressive duration split, near-critically damped spring-ratio defaults with a runtime translation table, enter/exit asymmetry, translate-distance caps, stagger budgets.

Rewrites marketing and product claims into the premium register – every quantified claim carries a number, a baseline, and a footnote, headlines run 7 words or fewer in sentence case with a concrete noun.

Diagnoses why a design reads as generated or template-built – scans the artifact against a 22-row anti-pattern table where every row names the tell, the perceptual mechanism behind it, and the skill that fixes it, then returns a severity-ranked.

Designs the lighting plan for a product shot – one motivated key sized 2-3x the product and placed close, one dedicated source per surface built in layers of 4-8, shaped gradient reflections on glossy faces, rim light 1-2 stops above the adjacent background.

Translates a product's materials into optical behavior – bright-field or dark-field schemes for glass, anisotropic highlight streaks aligned with brushed-metal grain, raking light at 10-30 degrees for leather and fabric.

Specifies camera and framing for product hero and detail shots – the 85-105mm focal band, sensor plane parallel to the product's dominant face, f/8-16 focus-stacked heroes vs f/2-4 single-element details, product filling 25-40 percent of frame.

Designs the stage behind and around a product – seamless sweeps held at 90-97 percent luminance, one-direction background gradients with the dark zone placed behind the product's bright edge, near-black stages for dark-field work.

Builds the complete campaign shot manifest for a product launch behind a hardware gate – physical products get the six-shot spine (three-quarter hero at 15-30 degrees above deck, straight-on elevation, top-down flat lay.

Directs photoreal 3D product renders – 0.5-1.5mm edge fillets on every mesh, 2-5 percent roughness variation via smudge and fingerprint maps, 2-4 percent grain over large gradients, physically correct glass at IOR 1.45-1.52.

Executes photoreal 3D product renders end to end in headless Blender: GPU detection (Metal/CUDA/OptiX/HIP) with CPU fallback, denoiser detection, a parametric device-mesh builder, and PBR node graphs wired from the direction skills' numbers.

Composites pixel-perfect UI onto device photos and renders – screen whites about 1 stop below the key highlights so displays never clip, a diagonal gradient reflection at 3-8 percent opacity with a 45-degree cut edge laid over the UI.

Routes each creative deliverable to the right AI image or video model by its failure mode – typography, vector, photoreal with brand hex, campaign consistency, dialogue, human motion – and compiles one creative spec into each model family's prompt dialect.

Writes photorealistic product and editorial image prompts using camera vocabulary as the control surface – mandatory focal length, aperture, shot type, one named directional light, material optics, and an imperfection budget.

Locks a multi-asset AI campaign to one visual identity using reference mechanisms instead of re-described prose – a frozen content-hashed set block, a curated style-reference code library, brand-style locks, and multi-image references.

Decides where words in AI-generated imagery get rendered – in-image with a typography-capable model only when the text is a short quoted string within the ceiling this skill's platform-facts block states, composited in the design layer otherwise.

Directs AI-generated video by decomposing a brief into single-intention shots – one camera move plus one subject action per 5 to 8 second clip – then serializing the shot list to the target model's multi-shot mechanism with timestamp-exact audio cues and…

Runs the selection and revision discipline that separates shipped AI assets from lucky ones – a minimum four-candidate set per final asset, a tell checklist covering rendered-text read-back, hands, logo geometry, shadow, depth-of-field.

Escalates AI image and video work from prompt-only to pipeline mode when fidelity is contractual – structure conditioning for layout-exact generation, hex color conditioning for brand palettes, pinned seeds and saved workflows for reproducibility.

Turns real product UI into marketing-grade screenshot stills – flat-on framing, 12-16px radius frame, 1px hairline border, 2x capture, 20-40 percent edge crop, theme matched to the page.

Decides when a physical device frame earns its place in a marketing still and directs the shot when it does – flat-on at 0 degrees or a precise 30-45 degree three-quarter, one shadow direction per frame, soft contact shadow at every support point.

Art-directs the website hero as a still – a headline pyramid with at most 2 CTAs, the product screenshot as the hero art at 16px radius cropped off the fold, exactly one atmosphere zone per page, and the asset budget that keeps the fold fast.

Designs App Store and Google Play screenshot sets to exact per-device store specs – with a runnable preflight validator, 3-6-word benefit-first caption bands, and a first-three sequence that reads as one pitch.

Builds the Product Hunt gallery as a five-beat visual story – 5-8 slides at 1270x760 under 3MB each with the roles hook, workflow, differentiator, proof, CTA – plus an animated 240x240 thumbnail designed to survive the roughly 80px feed render.

Produces the social still set – OG and link-card images at 1200x630, quote cards, and static ad images – recomposed per platform from one master composition and shipped as runnable HTML/CSS compositions.

Designs explanatory marketing graphics as production-ready specs – comparison graphics, stat tiles, feature diagrams, docs and blog artwork, and investor-deck visuals – one idea per graphic, the hero number set at about 4 times its caption size.

Directs a software launch trailer end to end – the hook-vignettes-crescendo-endcard spine, a BPM-derived beat grid mapped to the track's phrases and drop that quantizes every cut, montage acceleration into the climax, anticipation holds before reveals.

Choreographs device and app-window reveal shots for launch films – chassis settles from 0.92 to 1.0 scale, screen ignition delayed 100-200ms after the frame lands, a single timed light sweep to sell the material, then a slow camera push.

Animates recreated product UI for marketing films so the interface performs like a human is using it – cursors on curved paths with hover pauses, typing at jittered per-character speed, one staged state change per beat, constant micro camera drift.

Directs how a brand mark behaves in motion – the trailer end-card resolve, seamless ambient loops with position and velocity matched at the seam, a rest phase so the loop reads calm, and marks that survive a 240px animated thumbnail.

Designs marketing-site hero and launch-page motion – scroll-linked scrub sections pinned 100-200vh, one signature mechanic per page, small masked reveals, and hard performance gates – so the page choreographs the story without scroll-jacking, jank.

Builds the complete asset manifest for a software launch – every artifact the launch type demands (Product Hunt, app store, demo day, press), each with its spec, its producing skill, and its deadline class.

Structures launch-day announcement posts for X and LinkedIn to platform mechanics – a single long-form X post with a native clip and the link in a reply as the default (the multi-post thread is the escape hatch for genuinely serial stories).

Runs the community side of a software launch – the Show HN package (the title convention, the link-versus-text post decision, the author's first technical comment, and a criticism-response register) plus seeding etiquette for subreddits.

Runs the pre-launch audience build – a 4-8 week teaser arc with one escalating reveal per week, the Product Hunt Coming Soon page as the platform-native pre-commitment surface.

Assembles the self-serve press kit – one ungated press URL with a one-paragraph boilerplate, a fact sheet where every number carries a date and a source, founder bios with high-resolution headshots, an SVG plus PNG logo pack in light and dark modes.

Builds the accelerator demo-day showcase package to the one-minute compression standard – a spoken script capped at 150 words that opens with the strongest traction metric, one slide carrying 10 or fewer words at 40pt or larger on a white background.

Turns launch-day results into the second launch – a metrics checklist filled during the day itself (hourly rank, referrer split, signup count), a public day-plus-2 retrospective shipped within 48-72 hours as an X post and a LinkedIn document post.

Audits every brand surface – product, marketing site, social, email, decks, ads – against the brand's locked fixed core and computes a drift score with a per-surface conformance table and a fix routing plan.

Maps the visual conventions of a product category – competitor palettes, type voices, and layout codes plotted on a two-axis map – and scores whether a brand lands in an ownable empty quadrant or blends into category wallpaper.

Compiles brand guidelines prose into a three-layer token set in DTCG-format JSON – ready for Style Dictionary and Figma Variables import – and sorts every rule into a declared enforcement tier: build-lintable validators, CI-script checks.

Specifies the misuse gallery for a brand book – at least 6 wrongs per element class (mark, color, type, component, layout) with red strikes and corrective captions, emitted as a self-contained HTML misuse sheet where the wrongs are constructible in code.

Stress-tests a brand system at its worst cases before any hero application – 16px favicon, email footer, co-brand lockup, dark mode, single-color print, data-dense screen, legal document, tight social crop.

Reviews a single artifact – landing page, deck, ad, social image, launch gallery – against the brand's locked constants before it ships, returning a pass, fix-and-resubmit, or blocked verdict with the exact rule each violation breaks.

Scores any creative artifact – page, campaign asset, deck, video, brand application – against four verifiable quality bars (consumer, startup, premium SaaS, keynote), each a pass/fail checklist with numeric thresholds and per-artifact-type applicability.

Resolves contradictions between design rules with thirteen if/then arbitration decisions – pure black vs near-black, pill vs rectangle CTAs, duration ceilings, entrance-ease bans vs camera eases, scroll pinning vs scroll-jacking, dark vs light default.

Writes the art direction brief every production job starts from – a fixed core of 3-5 frozen constants, 3-5 designed flex variables, camera-level operational constraints instead of adjectives, reference anchors instead of re-described prose.

Reviews finished creative work – landing pages, campaign assets, trailers, decks, generated imagery – the way a creative director does: runs the swap, squint, mono, 16px, and strategy-word test battery with a required evidence artifact per test.

Audits a multi-asset campaign for drift against its frozen anchors – accent hex, canvas values, CTA radius, type weight set, motion damping-ratio floor, duration bands, shot grammar.

Runs the final ship-or-hold gate on any creative artifact – performance budgets, accessibility floors, smallest-surface passes, the honesty contract on screenshots and claims, and a sweep of the named anti-pattern tells.