メニュー
Build multi-shot narrative image, video, and audio workflows with `vg generate`. Use this for storyboards, shot lists, multi-prompt video, first-frame to last-frame pipelines, social stories, brand films, and sequence continuity.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use the `vg generate` CLI to search, inspect, run, and manage 1200+ generative model endpoints. Trigger when the user asks to "generate an image", "make a video", "search models", "run a model", "fetch schema", "check pricing", "upload an asset", "queue async job", "track request", or any direct interaction with the model endpoint catalog. This is the foundational skill. Every other media skill in this repo executes its work through `vg generate` commands. Use `--json` whenever the output will be parsed by an agent.
Build 3D browser apps AND games with Three.js (r150+, ES modules): scene setup, geometries, materials, lighting, animation, GLTF models, physics (Rapier/cannon-es/arcade), character controllers, follow/third-person cameras, fixed-timestep loops, post-processing, and performance/debugging. Use for 'create a three.js scene/app/showcase', any 3D web content, and turning a scene into a game — 'add physics to my three.js game', 'third-person/character controller', 'follow camera', 'jump and gravity', 'collide with / pick up objects', 'structure my 3d game'. Trigger: threejs, three.js, 3D scene, WebGL scene, GLTF/GLB, OrbitControls, Rapier, cannon-es, 3D game, third-person controller. For 2D games use `phaser`; for engine-agnostic feel/balance/level craft use the game-craft skills.
Turn a character anchor into an engine-loadable animated spritesheet by generating ONE labeled pose-board image (a grid of the same character in the frames of an action) and slicing it. Works for any action — idle, run, jump, attack, hurt, crouch, death, roll. Generates a per-frame-labeled pose board on a flat chroma matte, recovers/slices the frames, keys + despills, snaps to crisp native pixels, normalizes with headroom, and packs spritesheet.png + a manifest. Triggers: 'sprite animation', 'animated spritesheet', 'attack animation', 'walk/run cycle', 'animate this character', 'game sprite animation', 'sprite pose sheet'.
Interview the user relentlessly about the game they want to make until the idea is build-ready, resolving each branch of the design tree one question at a time. Use when the user has a fuzzy or one-line game idea and wants to flesh it out before building: 'grill me about my game', 'help me figure out what game to make', 'stress-test my game idea', 'interview me about the game', 'I want to make a game but I'm not sure what'. Not for when the user already knows what they want built — go straight to game-playbook.
Build and ship Three.js apps on Capacitor iOS with Vite and Swift Package Manager: GLTF loading, assets_index animation UI, OrbitControls mouse/touch mappings, and iOS sync/run troubleshooting.
Build 2D browser games with Phaser 4: scene lifecycle, sprites, Arcade/Matter physics, tilemaps, WebGL rendering, filters, lighting, shaders, DynamicTexture/RenderTexture, SpriteGPULayer, TilemapGPULayer, and Phaser 3 to 4 migration. Trigger: phaser, phaser 4, phaser v4, create phaser game, add phaser scene, phaser sprite, phaser physics, phaser tilemap, phaser arcade, phaser webgl renderer, phaser filters, phaser SpriteGPULayer, phaser TilemapGPULayer, migrate phaser 3 to 4.
| name | storytelling |
| description | Build multi-shot narrative image, video, and audio workflows with `vg generate`. Use this for storyboards, shot lists, multi-prompt video, first-frame to last-frame pipelines, social stories, brand films, and sequence continuity. |
vg generateUse this skill when the user wants a sequence, not a single asset. Load references as needed:
references/shot-planning.mdreferences/workflows.mdreferences/examples.mdLoad model-catalog alongside this skill for default endpoint choices.
The goal is to produce clear story beats and executable vg generate runs. Avoid generic inspiration copy, fake dialogue, and em dashes.
Ask only when missing information affects execution.
model-catalog defaults unless the user names one.Follow the standard workflow in the generate skill: resolve endpoint → inspect schema/pricing → upload references (first frame, character, product, voiceover) → run each shot --async + status → download to ./outputs/story/{request_id}_{index}.{ext}. Default endpoint IDs are in the Model routing section below.
Then choose the sequence route:
model-catalog.model-catalog.Return a shot table with endpoint, request id, prompt summary, local path, and any continuity issues. Genmedia downloads clips; it does not replace a timeline editor unless the chosen model returns a complete stitched video.
Plan every sequence as beats first:
For each shot, write:
Use this structure for each shot:
SHOT [number], [duration]:
[story purpose]. [subject and action]. [location and time]. [camera framing].
[camera movement]. [lighting and color]. [continuity anchor]. [transition or
relationship to previous shot].
Keep one shot to one clear action unless the selected model supports multi-shot or timeline prompting.
bytedance/seedance-2.0/text-to-video,
bytedance/seedance-2.0/image-to-video, or
bytedance/seedance-2.0/reference-to-video.xai/grok-imagine-video/text-to-video or
xai/grok-imagine-video/image-to-video.fal-ai/kling-video/v3/pro/text-to-video, then
fal-ai/kling-video/v3/pro/image-to-video, then
alibaba/happy-horse/text-to-video or
alibaba/happy-horse/image-to-video.openai/gpt-image-2 at quality=high.veed/fabric-1.0, veed/fabric-1.0/text, or fal-ai/creatify/aurora.Before returning: