weshopai
OrganizationCreative AI Skills for Codex, Claude Code, Cursor, Deepseek harness and any Agent Skills-compatible runtime.
Categories
Indexed Skills (117)
3d-animation-short-generator
Create one complete story-driven stylized 3D animated short from premise through original character and environment references, continuity-safe multi-shot planning, rendered clips, assembly, and final film QC. Use when the requested deliverable is a finished narrative animation with multiple causally linked shots. Unlike plan-film-storyboard (relationship 0.78), this owns generated assets and final media; use plan-film-storyboard for planning-only delivery. Unlike generate-video (0.62), this owns a complete multi-shot story; use generate-video for one standalone shot. It can consume develop-story (0.55), call generate-video per shot, and hand accepted clips to combine-videos (0.38). Do not use for product-benefit ads, supplied-footage editing, one still, one clip, or imitation of a named studio or living artist.
ai-banner-design
Research and create a channel-ready commercial banner for a specific placement such as a website or SaaS hero, YouTube channel banner, paid display ad, social cover, email header, marketplace campaign, or event promotion. Use when current dimensions, safe areas, crop behavior, file requirements, industry conventions, references, exact copy, and a distinctive banner composition must be resolved before image generation; do not use for posters, thumbnails, responsive web implementation, multipage campaigns, or generic images without a banner placement.
ai-product
Place one supplied real product into a new commercial studio, material, lifestyle, or campaign-ready scene while preserving product silhouette, proportions, construction, materials, colors, packaging, labels, logos, controls, and product count. Use for AI product photography, product background replacement, still-life scenes, lifestyle product images, material backdrops, and copy-space variants; do not use for white-background mockups, inventing a new product, packaging design, posters, product-detail layouts, collages, or background removal.
brand-promo-video-generator
Create one complete short promotional video for a brand, app, website, shop, service, launch, or mixed brand system from verified identity assets, a brand-truth and provenance manifest, campaign direction, timed beats, generated media, assembly, and final authenticity QC. Use when the campaign subject is broader than one product-benefit claim. Unlike make-product-commercial (relationship 0.86), choose this for brand-system, digital-product, service, shop, or launch storytelling; choose make-product-commercial when one supplied product and its truthful causal benefit own a 5–30 second ad. Unlike make-explainer-video (0.58), this optimizes promotion and CTA rather than source-backed education. It may call generate-video (0.49) for shots and combine-videos (0.36) for assembly. Do not use for one generic shot, editing-only supplied footage, invented claims, or approximate identity assets.
correct-video-color
Correct white balance, exposure, contrast, and shot-to-shot color consistency in supplied video without changing its content or intended creative look. Use for technical color correction and clip matching; do not use for stylized restyling, adding effects, repairing objects, or editing the sequence.
create-animal
Generate one specified real species or original animal creature in an anatomically and ecologically coherent scene. Use for standalone animal creation from a brief; do not use for a portrait of the user's supplied pet, a humanoid character, an NPC, or animating an existing animal image.
create-character
Create an original reusable character through a canonical-first workflow. Generate and review one polished multi-panel character design sheet first; only after the user confirms expansion, create seven separate reference-bound assets for front, back, close-up, lighting, final look, and two scenes. Use for recurring characters in comics, storyboards, images, or video; also owns requests previously phrased as character reference sheets. Do not use for game-role-specific NPC design, small avatars, fashion lookbooks, or changing an existing person's appearance.
create-custom-skill
Create, revise, check, and locally install one user-owned Agent Skill from a new brief, the current conversation, or an existing local Skill. Use when the user asks to make, save, capture, customize, validate, or update their own reusable local Skill. This Skill owns the complete local lifecycle; official package submission is a separate maintainer intake and review workflow and is never implied by local creation.
create-image-deck
Create a consistent image-first presentation deck from supplied content, data, and authorized references, with a slide manifest, one generated visual per page, exact-content checks, and an inspected deck package. Use for visual explainers, image-led pitch or teaching decks, NotebookLM-like illustrated presentations, or full-bleed slide stories; do not use for a single poster, social carousel, ordinary text-first office slides, unsupported factual research, or automatic publication.
create-mascot-logo
Create a highly simplified square IP mascot logo through three product-relevant directions and separately generated candidates. Use for cute rounded animal, creature, robot, plant, or object mascots meant to work as a compact brand symbol; do not use for wordmarks, monograms, general logo lockups, detailed character illustrations, avatars, or character reference sheets.
create-social-carousel
Plan and create one ordered, visually consistent multi-page social carousel with mobile-readable copy, a reusable visual master, and page-by-page acceptance. Use for Xiaohongshu, Instagram, LinkedIn, or similar swipeable educational and promotional posts; do not use for one poster, one thumbnail, unordered photo collages, product-detail pages, or automatic publishing.
develop-story
Develop a story idea, weak premise, outline, or draft into a structured foundation with premise options, story core, ending direction, protagonist engine, world pressure, causal beats, scene briefs, and prioritized revision. Use before adapting a general story into comics, film, short drama, or prose; do not use for final screenplay formatting, complete novel prose by default, film storyboards, comic-page planning, or media generation.
expand-image
Extend one supplied image to a new aspect ratio or larger canvas by generating only the added surroundings while preserving the existing subject, scene, copy, and composition. Use for outpainting, portrait-to-landscape or square-to-vertical conversion, platform-size adaptation, safe-area or copy-space creation, and fixing ratio mismatches after another image Skill; do not use for simple upscaling, ordinary cropping, background replacement, redesign, or adding new story content.
add-braces
Add one realistic braces treatment to the visible teeth of an authorized smiling portrait while preserving identity, teeth layout, lips, bite, face, expression, skin, hair, clothing, and background. Use for cosmetic orthodontic visualization of metal, ceramic, or colored brackets; do not use for diagnosis or treatment prediction.
add-speech-bubble
Add position-accurate speech or thought bubbles with exact text to a supplied image. Use for comics, reaction images, dialogue scenes, or annotations where bubble shape, tail target, stacking, and text legibility matter; do not use for translating existing image text, poster typography, or general caption layout.
add-tattoo
Place one specified tattoo design on visible skin in an authorized image while preserving identity, anatomy, skin tone and texture, clothing, pose, lighting, and background. Use for placement, scale, orientation, blackwork, linework, or color tattoo previews; do not use to copy a living artist's protected design or claim permanent-result accuracy.
add-video-effect
Add one specified visual effect to supplied video at a defined time and region while preserving the underlying footage. Use for localized glow, particles, weather, light, impact, or transition effects; do not use for whole-video restyling, color correction, footage assembly, or removing overlays.
animate-image
Animate one supplied still into a short video with bounded subject and camera motion while preserving identity and composition. Use for image-to-video motion from a specific source frame; do not use for text-only video generation, talking portraits, full video restyling, or clip editing.
apply-makeup
Apply one makeup look to an authorized portrait while preserving identity, facial structure, natural skin texture, hair, clothing, pose, lighting, and background. Use for beauty, bridal, editorial, stage, product-shade, or reference-makeup previews; do not use to reshape the face or change skin identity.
apply-photo-filter
Apply one controlled photographic color-and-texture treatment while preserving subjects, geometry, identity, and scene. Use for film emulation, monochrome, faded, warm, cool, matte, grain, or reference-led grading; not content transformation.
change-bangs
Add or change only bangs in one authorized portrait while preserving identity, the rest of the hairstyle, hair color, face, clothing, pose, and background. Use for curtain, blunt, wispy, side-swept, micro, or reference-fringe previews; do not use for a full haircut replacement.
character-reference-sheet
Compatibility entry for older character-sheet requests. Route the complete request to create-character, which now owns canonical sheet creation and the full eight-image character production pack. Do not submit an independent generation run from this Skill.
clean-room
Remove specified clutter from a room photograph with an AI-native contextual edit while preserving architecture, fixed furniture, perspective, materials, lighting, and the room's identity. Use for real-estate, hospitality, or interior cleanup, not redesign or restyling.
colorize-image
Add plausible, context-aware color to a black-and-white or grayscale image while preserving identity, content, tonal structure, texture, era cues, and damage. Use for archival-style, family, editorial, or creative colorization; disclose when colors are inferred rather than documented.
combine-videos
Combine supplied clips into one ordered video with intentional trims, transitions and audio handling; do not regenerate footage.
compose-lookbook
Arrange several supplied outfit photographs into one intentional lookbook page with an editorial grid, hierarchy, spacing, and optional exact copy. Use after outfit images exist; do not invent outfits or collapse all looks into one person.
create-avatar
Create one small-size-readable profile avatar from an authorized portrait or fictional brief. Use for account icons, community profiles, and compact identity marks; do not use for professional headshots, full character sheets, ID photos, or full-body character artwork.
create-cad
Create and validate one dimensioned parametric CAD part, returning STEP as the primary artifact rather than a render-only image. Use for manufacturable mechanical geometry with explicit dimensions and constraints; do not use for concept renders, floor plans, architecture, diagrams, or artistic 3D imagery.
create-flag
Design one original symbolically coherent flag that remains recognizable at distance and in flat form. Use for fictional places, organizations, events, or identity systems needing vexillographic design; do not use for logos, banners, posters, existing national-flag alteration, or manufacturing specifications.
create-logo
Design one selected original logo mark or lockup from a brand brief after concept reduction and application checks. Use for a core brand symbol, wordmark, monogram, or compact lockup; do not use for full brand-identity systems, posters, packaging boards, flags, or generic illustrations.
create-npc
Create one original non-player character image whose occupation, faction, world function, and gameplay role are visually legible. Use for game-ready NPC concepts tied to a setting and role; do not use for generic standalone characters, multi-view reference sheets, player avatars, or portraits of real people.
edit-social-video
Turn supplied footage into one platform-ready social video using current placement rules, a strong opening, pacing, and evidence-faithful captions. Use for TikTok, Reels, Shorts, or feed edits assembled from existing media; do not use for generating all footage, combining clips without a social brief, podcast programs, or simple color correction.
eye-color-change
Change only both iris colors in one authorized portrait while preserving identity, pupil size, sclera, catchlights, eyelids, gaze, facial structure, skin, hair, clothing, lighting, and background. Use for natural contact-lens or eye-color previews; do not use for eye-shape changes or medical simulation.
fashion-model-replacement
Replace the person or mannequin in an already-dressed fashion image while preserving the complete outfit, garment construction, fit cues, branding, pose, crop, and requested scene. Use when a user asks for a different fashion model, mannequin-to-human conversion, market-specific adult model, text-described model, or supplied target-person reference; do not use for putting a separate garment on a person, changing only pose, inventing an outfit, face-only entertainment swaps, or general photoshoots.
generate-video
Generate one coherent video from a text or multimodal shot brief using a deliberately selected current model and one observable action. Use for new cinematic footage when no narrower video Atom owns the outcome; do not use for animating one supplied still, editing supplied footage, talking portraits, intros, or multi-clip assembly.
id-photo-format
Prepare one authorized portrait for one explicitly named ID-photo or passport-photo specification using current issuing-authority research, neutral capture correction, measured crown-to-chin cropping, exact dimensions, background, format, and DPI. Use for passport, visa, license, school, application, or print preparation; never assume universal ear, crop, lighting, or AI-edit rules and never fabricate identity.
image-combiner
Merge two to five supplied images into one newly generated, visually coherent image while preserving the requested subjects and reconciling scale, perspective, lighting, color, depth, and occlusion. Use when elements from separate photos should naturally coexist in one scene; do not use for side-by-side collages, poster layouts, background replacement only, local retouching, simple overlays, or pixel-exact compositing.
make-beat-synced-video
Edit supplied video clips or stills to supplied music using measured beat and energy analysis, an inspectable edit decision list, restrained beat-timed effects, deterministic rendering, and full-export QC. Use for music montages, beat-cut reels, rhythm edits, photo montages, and platform-format variants; do not use for generating new footage, composing music, dialogue-led narrative editing, arbitrary effect spam, or downloading unlicensed media.
make-explainer-video
Research, script, plan, generate, and assemble one short narrated explainer video with traceable facts, beat-timed visuals, subtitles, and intelligible audio. Use for educational mini-documentaries, topic explainers, report-to-video, and narrated video essays; do not use for imitating a named publisher's house style, unsupported standalone voice/music generation, product commercials, talking portraits, or simple clip assembly.
make-kinetic-typography
Create one polished, accessible kinetic-typography animation from supplied copy and brand direction, with readable static typography, intentional split/reveal motion, deterministic frame review, and a runnable or rendered deliverable. Use for animated headlines, title cards, lyric or caption-led pieces, variable-font motion, and text-on-path animation; do not use for ordinary subtitles, a full product promo, static posters, inaccessible character splitting, or unlicensed font use.
make-music-video
Analyze one authorized song and supplied or rights-cleared footage, design a lyric- and music-aware edit decision list, render one synchronized music video, and inspect the final cut. Use for montage, performance, narrative, or archival music-video edits from existing media; do not use for downloading copyrighted sources, generating a song, editing dialogue-led podcasts, simple clip concatenation, or adding one isolated effect.
casting-cn
Create a Chinese screen-role casting package from a six-field role card: one pure-white, face-first casting portrait for approval, then only selected identity-bound full-body, expression-grid, or turnaround assets. Use for short-drama, film, and casting-character roles; unlike `create-character` (relationship 0.84), choose this for actor-style face selection and a confirmation gate, and choose `create-character` for a canonical multi-panel production pack and scenes. Do not use for real-person edits, game-role NPC concepts, or generic character sheets.
3d-animation-short-generator
Create one complete story-driven stylized 3D animated short from premise through original character and environment references, continuity-safe multi-shot planning, rendered clips, assembly, and final film QC. Use when the requested deliverable is a finished narrative animation with multiple causally linked shots. Unlike plan-film-storyboard (relationship 0.78), this owns generated assets and final media; use plan-film-storyboard for planning-only delivery. Unlike generate-video (0.62), this owns a complete multi-shot story; use generate-video for one standalone shot. It can consume develop-story (0.55), call generate-video per shot, and hand accepted clips to combine-videos (0.38). Do not use for product-benefit ads, supplied-footage editing, one still, one clip, or imitation of a named studio or living artist.
ai-banner-design
Research and create a channel-ready commercial banner for a specific placement such as a website or SaaS hero, YouTube channel banner, paid display ad, social cover, email header, marketplace campaign, or event promotion. Use when current dimensions, safe areas, crop behavior, file requirements, industry conventions, references, exact copy, and a distinctive banner composition must be resolved before image generation; do not use for posters, thumbnails, responsive web implementation, multipage campaigns, or generic images without a banner placement.
ai-product
Place one supplied real product into a new commercial studio, material, lifestyle, or campaign-ready scene while preserving product silhouette, proportions, construction, materials, colors, packaging, labels, logos, controls, and product count. Use for AI product photography, product background replacement, still-life scenes, lifestyle product images, material backdrops, and copy-space variants; do not use for white-background mockups, inventing a new product, packaging design, posters, product-detail layouts, collages, or background removal.
backrooms-dreamcore
Create one personalized Backrooms Level-0/dreamcore space dossier and, when requested, a reference-led photoreal still or short shot from a supplied character, pet, object, or scene image. Use for “my Backrooms”, image-specific liminal-space transformations, and dossier-first dreamcore work; unlike `restyle-room` (relationship 0.58), choose this when Level-0 grammar and the personalized space interpretation matter; unlike `generate-video` (0.66), choose it when the reference, spatial tendencies, and dossier own the result. Do not use for lore Q&A, generic horror imagery, psychological assessment, or ordinary room restyling.
brand-promo-video-generator
Create one complete short promotional video for a brand, app, website, shop, service, launch, or mixed brand system from verified identity assets, a brand-truth and provenance manifest, campaign direction, timed beats, generated media, assembly, and final authenticity QC. Use when the campaign subject is broader than one product-benefit claim. Unlike make-product-commercial (relationship 0.86), choose this for brand-system, digital-product, service, shop, or launch storytelling; choose make-product-commercial when one supplied product and its truthful causal benefit own a 5–30 second ad. Unlike make-explainer-video (0.58), this optimizes promotion and CTA rather than source-backed education. It may call generate-video (0.49) for shots and combine-videos (0.36) for assembly. Do not use for one generic shot, editing-only supplied footage, invented claims, or approximate identity assets.
casting
Create a Chinese screen-role casting package from a six-field role card: one pure-white, face-first casting portrait for approval, then only selected identity-bound full-body, expression-grid, or turnaround assets. Use for short-drama, film, and casting-character roles; unlike `create-character` (relationship 0.84), choose this for actor-style face selection and a confirmation gate, and choose `create-character` for a canonical multi-panel production pack and scenes. Do not use for real-person edits, game-role NPC concepts, or generic character sheets.
change-pose
Change the pose of one already-dressed person while preserving identity, body proportions, complete outfit, branding, accessories, background, lighting, and commercial image continuity. Use for fashion reposing, pose variants, catalog stance changes, walking or seated poses, and pose-reference requests; do not use for changing the model, changing clothing, creating a new photoshoot, animating video, or editing only an expression or hand.
character-reference-sheet
Compatibility entry for older character-sheet requests. Route the complete request to create-character, which now owns canonical sheet creation and the full eight-image character production pack. Do not submit an independent generation run from this Skill.
character-toy-grid-motion
Use for an uploaded character transformed into an identity-preserved 9:16 designer-toy poster with six irregular rounded panels and a motion version that opens on one panel before returning to the full grid. Unlike character-reference-sheet (relationship 0.73), choose this for the toy-grid motion deliverable; choose character-reference-sheet for broad design reference. Unlike generate-video (0.61), choose this when the six-panel layout is mandatory.
correct-video-color
Correct white balance, exposure, contrast, and shot-to-shot color consistency in supplied video without changing its content or intended creative look. Use for technical color correction and clip matching; do not use for stylized restyling, adding effects, repairing objects, or editing the sequence.
create-animal
Generate one specified real species or original animal creature in an anatomically and ecologically coherent scene. Use for standalone animal creation from a brief; do not use for a portrait of the user's supplied pet, a humanoid character, an NPC, or animating an existing animal image.
create-character
Create an original reusable character through a canonical-first workflow. Generate and review one polished multi-panel character design sheet first; only after the user confirms expansion, create seven separate reference-bound assets for front, back, close-up, lighting, final look, and two scenes. Use for recurring characters in comics, storyboards, images, or video; also owns requests previously phrased as character reference sheets. Do not use for game-role-specific NPC design, small avatars, fashion lookbooks, or changing an existing person's appearance.
create-image-deck
Create a consistent image-first presentation deck from supplied content, data, and supplied references, with a slide manifest, one generated visual per page, exact-content checks, and an inspected deck package. Use for visual explainers, image-led pitch or teaching decks, NotebookLM-like illustrated presentations, or full-bleed slide stories; do not use for a single poster, social carousel, ordinary text-first office slides, unsupported factual research, or generic publication planning.
create-mascot-logo
Create a highly simplified square IP mascot logo through three product-relevant directions and separately generated candidates. Use for cute rounded animal, creature, robot, plant, or object mascots meant to work as a compact brand symbol; do not use for wordmarks, monograms, general logo lockups, detailed character illustrations, avatars, or character reference sheets.
create-social-carousel
Plan and create one ordered, visually consistent multi-page social carousel with mobile-readable copy, a reusable visual master, and page-by-page acceptance. Use for Xiaohongshu, Instagram, LinkedIn, or similar swipeable educational and promotional posts; do not use for one poster, one thumbnail, unordered photo collages, product-detail pages, or automatic publishing.
create-white-background-product-mockup
Create one clean, product-faithful mockup on a plain white background from a description or supplied reference image. Use for catalog packshots, apparel flat lays, ghost-mannequin presentations, sketch-to-garment mockups, or handbag, shoe, apparel, packaging, and object mockups where geometry, material, color, branding, and visible text must not drift; do not use for lifestyle scenes, decorative commercial backgrounds, transparent cutouts, packaging concept design, or posters.
data-driven-ad-creative-video
Use for controlled ad-video variants from approved tabular data from approved user facts and assets; unlike $generate-video (relationship 0.67), choose this when the stated specialized output and preservation contract are required; choose $generate-video for its generic final-media contract.
expand-image
Extend one supplied image to a new aspect ratio or larger canvas by generating only the added surroundings while preserving the existing subject, scene, copy, and composition. Use for outpainting, portrait-to-landscape or square-to-vertical conversion, platform-size adaptation, safe-area or copy-space creation, and fixing ratio mismatches after another image Skill; do not use for simple upscaling, ordinary cropping, background replacement, redesign, or adding new story content.
article-handdrawn-illustrations
Turn a supplied article or one explicit idea into sparse hand-drawn explanatory body illustrations with fresh physical metaphors and a deadpan black character performing the core action. Use for Chinese article illustrations, body-image shot lists, conceptual hand-drawn explainers, or editing an accepted image in this visual system; do not use for sourced infographics, social carousels, slide decks, posters, logos, generic illustration, or generic publication planning.
cinematic-video-prompt
Use for cinematic or image-to-video motion prompt from approved user facts and assets; unlike $generate-video (relationship 0.67), choose this when the stated specialized output and preservation contract are required; choose $generate-video for its generic final-media contract.
create-custom-skill
Create, revise, check, and locally install one user-owned Agent Skill from a new brief, the current conversation, or an existing local Skill. Use when the user asks to make, save, capture, customize, validate, or update their own reusable local Skill. This Skill owns the complete local lifecycle; official package submission is a separate maintainer intake and review workflow and is never implied by local creation.
cross-model-video-prompt-adapter
Use for semantic adaptation of an existing video prompt from approved user facts and assets; unlike $generate-video (relationship 0.67), choose this when the stated specialized output and preservation contract are required; choose $generate-video for its generic final-media contract.
develop-story
Develop a story idea, weak premise, outline, or draft into a structured foundation with premise options, story core, ending direction, protagonist engine, world pressure, causal beats, scene briefs, and prioritized revision. Use before adapting a general story into comics, film, short drama, or prose; do not use for final screenplay formatting, complete novel prose by default, film storyboards, comic-page planning, or media generation.
change-pose
Change the pose of one already-dressed person while preserving identity, body proportions, complete outfit, branding, accessories, background, lighting, and commercial image continuity. Use for fashion reposing, pose variants, catalog stance changes, walking or seated poses, and pose-reference requests; do not use for changing the model, changing clothing, creating a new photoshoot, animating video, or editing only an expression or hand.
create-white-background-product-mockup
Create one clean, product-faithful mockup on a plain white background from a description or supplied reference image. Use for catalog packshots, apparel flat lays, ghost-mannequin presentations, sketch-to-garment mockups, or handbag, shoe, apparel, packaging, and object mockups where geometry, material, color, branding, and visible text must not drift; do not use for lifestyle scenes, decorative commercial backgrounds, transparent cutouts, packaging concept design, or posters.
casting
Create a Chinese screen-role casting package from a six-field role card: one pure-white, face-first casting portrait for approval, then only selected identity-bound full-body, expression-grid, or turnaround assets. Use for short-drama, film, and casting-character roles; unlike `create-character` (relationship 0.84), choose this for actor-style face selection and a confirmation gate, and choose `create-character` for a canonical multi-panel production pack and scenes. Do not use for real-person edits, game-role NPC concepts, or generic character sheets.
backrooms-dreamcore
Create one personalized Backrooms Level-0/dreamcore space dossier and, when requested, a reference-led photoreal still or short shot from an authorized character, pet, object, or scene image. Use for “my Backrooms”, image-specific liminal-space transformations, and dossier-first dreamcore work; unlike `restyle-room` (relationship 0.58), choose this when Level-0 grammar and the personalized space interpretation matter; unlike `generate-video` (0.66), choose it when the reference, spatial tendencies, and dossier own the result. Do not use for lore Q&A, generic horror imagery, psychological assessment, or ordinary room restyling.
character-toy-grid-motion
Use for an uploaded character transformed into an identity-preserved 9:16 designer-toy poster with six irregular rounded panels and a motion version that opens on one panel before returning to the full grid. Unlike character-reference-sheet (relationship 0.73), choose this for the toy-grid motion deliverable; choose character-reference-sheet for broad design reference. Unlike generate-video (0.61), choose this when the six-panel layout is mandatory.
hair-color-change
Change only the hair color in one authorized portrait while preserving identity, hairstyle geometry, face, skin, clothing, pose, lighting, and background. Use for salon color previews, dye simulations, root-to-tip recoloring, highlights, balayage, or natural fantasy-color previews; do not use to change the haircut or add hair.
hairstyle-change
Replace only the hairstyle in one authorized portrait while preserving identity, face, requested hair color, body, clothing, pose, lighting, and background. Use for haircut, length, part, curl-pattern, updo, or reference-hairstyle previews; do not use for color-only edits, bangs-only edits, or bald previews.
make-birthday-video
Create one complete birthday memory video from authorized photos or clips, exact messages, and licensed or original audio. Use for birthday-specific greetings and memory montages; do not use for generic social edits, weddings, holidays, podcast programs, or generating an unrelated cinematic shot.
make-flowchart
Create one accurate readable flowchart from an explicit process model with verified nodes, decisions, labels, and connections. Use for process, decision, system, or workflow diagrams; do not use for infographics, floor plans, freeform illustrations, project schedules, or diagrams whose source logic is unknown.
make-holiday-card
Create one personalized holiday greeting card from authorized photos and exact copy while preserving identities and print-safe layout. Use for a single seasonal or holiday card; do not use for event posters, wedding portraits, birthday videos, generic photo collages, or multipage invitations.
make-infographic
Research or verify factual inputs, write a concise source-backed brief, and turn it into one readable infographic. Use when facts or data must be checked before visual design; never invent statistics or let image generation rewrite approved copy.
make-mugshot-photo
Create one clearly fictional or novelty mugshot-style portrait that cannot be mistaken for an official booking record.
brand-stream-mg
Create one logo-led neon brand motion from supplied inputs. Use when the distinct result and its preservation contract matter; unlike brand-promo-video-generator (relationship 0.68), choose this when logo-led neon brand motion is requested; choose brand-promo-video-generator for its narrower existing outcome. Do not use for generic media, source-provider tooling, generic publication planning, or unverified capabilities.
co-op-game-intro-generator
Create one two-player game menu and opening animation from supplied inputs. Use when the distinct result and its preservation contract matter; unlike generate-video (relationship 0.68), choose this when two-player game menu and opening animation is requested; choose generate-video for its narrower existing outcome. Do not use for generic media, source-provider tooling, generic publication planning, or unverified capabilities.
digital-product-promo-generator
Create one authentic digital-product promo plan and video from supplied inputs. Use when the distinct result and its preservation contract matter; unlike brand-promo-video-generator (relationship 0.68), choose this when authentic digital-product promo plan and video is requested; choose brand-promo-video-generator for its narrower existing outcome. Do not use for generic media, source-provider tooling, generic publication planning, or unverified capabilities.
dodge-game-video-generator
Create one character-led dodge-game reaction short from supplied inputs. Use when the distinct result and its preservation contract matter; unlike generate-video (relationship 0.68), choose this when character-led dodge-game reaction short is requested; choose generate-video for its narrower existing outcome. Do not use for generic media, source-provider tooling, generic publication planning, or unverified capabilities.
dot-matrix-brand-wordmark-motion
Create one exact brand wordmark dot-matrix motion from supplied inputs. Use when the distinct result and its preservation contract matter; unlike make-kinetic-typography (relationship 0.68), choose this when exact brand wordmark dot-matrix motion is requested; choose make-kinetic-typography for its narrower existing outcome. Do not use for generic media, source-provider tooling, generic publication planning, or unverified capabilities.
apply-makeup
Apply one makeup look to a supplied portrait while preserving identity, facial structure, natural skin texture, hair, clothing, pose, lighting, and background. Use for beauty, bridal, editorial, stage, product-shade, or reference-makeup previews; do not use to reshape the face or change skin identity.
add-braces
Add one realistic braces treatment to the visible teeth of a supplied smiling portrait while preserving identity, teeth layout, lips, bite, face, expression, skin, hair, clothing, and background. Use for cosmetic orthodontic visualization of metal, ceramic, or colored brackets; do not use for diagnosis or treatment prediction.
add-speech-bubble
Add position-accurate speech or thought bubbles with exact text to a supplied image. Use for comics, reaction images, dialogue scenes, or annotations where bubble shape, tail target, stacking, and text legibility matter; do not use for translating existing image text, poster typography, or general caption layout.
add-tattoo
Place one specified tattoo design on visible skin in a supplied image while preserving identity, anatomy, skin tone and texture, clothing, pose, lighting, and background. Use for placement, scale, orientation, blackwork, linework, or color tattoo previews; do not use to copy a living artist's protected design or claim permanent-result accuracy.
add-video-effect
Add one specified visual effect to supplied video at a defined time and region while preserving the underlying footage. Use for localized glow, particles, weather, light, impact, or transition effects; do not use for whole-video restyling, color correction, footage assembly, or removing overlays.
animate-image
Animate one supplied still into a short video with bounded subject and camera motion while preserving identity and composition. Use for image-to-video motion from a specific source frame; do not use for text-only video generation, talking portraits, full video restyling, or clip editing.
apply-photo-filter
Apply one controlled photographic color-and-texture treatment while preserving subjects, geometry, identity, and scene. Use for film emulation, monochrome, faded, warm, cool, matte, grain, or reference-led grading; not content transformation.
change-bangs
Add or change only bangs in one supplied portrait while preserving identity, the rest of the hairstyle, hair color, face, clothing, pose, and background. Use for curtain, blunt, wispy, side-swept, micro, or reference-fringe previews; do not use for a full haircut replacement.
children-english-story-episode
Use for an original, child-safe English-learning story episode with recurring characters, reusable scene assets, a target expression, short vertical story beats, and an optional song outline. Unlike character-reference-sheet (relationship 0.69), choose this when the deliverable is a complete learning episode; choose character-reference-sheet for only design views. Do not use source characters, source worlds, studio-style imitation, stored voice IDs, or automatic song generation.
clean-room
Remove specified clutter from a room photograph with an AI-native contextual edit while preserving architecture, fixed furniture, perspective, materials, lighting, and the room's identity. Use for real-estate, hospitality, or interior cleanup, not redesign or restyling.
colorize-image
Add plausible, context-aware color to a black-and-white or grayscale image while preserving identity, content, tonal structure, texture, era cues, and damage. Use for archival-style, family, editorial, or creative colorization; disclose when colors are inferred rather than documented.
combine-videos
Combine supplied clips into one ordered video with intentional trims, transitions and audio handling; do not regenerate footage.
compose-lookbook
Arrange several supplied outfit photographs into one intentional lookbook page with an editorial grid, hierarchy, spacing, and optional exact copy. Use after outfit images exist; do not invent outfits or collapse all looks into one person.
create-avatar
Create one small-size-readable profile avatar from a supplied portrait or fictional brief. Use for account icons, community profiles, and compact identity marks; do not use for professional headshots, full character sheets, ID photos, or full-body character artwork.
create-cad
Create and validate one dimensioned parametric CAD part, returning STEP as the primary artifact rather than a render-only image. Use for manufacturable mechanical geometry with explicit dimensions and constraints; do not use for concept renders, floor plans, architecture, diagrams, or artistic 3D imagery.
create-flag
Design one original symbolically coherent flag that remains recognizable at distance and in flat form. Use for fictional places, organizations, events, or identity systems needing vexillographic design; do not use for logos, banners, posters, existing national-flag alteration, or manufacturing specifications.
create-logo
Design one selected original logo mark or lockup from a brand brief after concept reduction and application checks. Use for a core brand symbol, wordmark, monogram, or compact lockup; do not use for full brand-identity systems, posters, packaging boards, flags, or generic illustrations.
create-npc
Create one original non-player character image whose occupation, faction, world function, and gameplay role are visually legible. Use for game-ready NPC concepts tied to a setting and role; do not use for generic standalone characters, multi-view reference sheets, player avatars, or portraits of real people.
Showing top 100 of 117 skills by quality score. Browse the full catalog at /browse.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.