@docyrus/animaguru 0.16.1 → 0.17.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@docyrus/animaguru",
-  "version": "0.16.1",
+  "version": "0.17.0",
   "description": "Agent-friendly AnimaGuru CLI built with CrustJS",
   "type": "module",
   "bin": {

package/skills/animaguru/SKILL.md

@@ -9,6 +9,8 @@ Use the `animaguru` CLI to generate media assets, create video workspaces, and s
 
 ## Discover and authenticate
 
+CLI 0.16.2 and newer require an update before work when a newer release is detected. If a command exits with `cli_update_required`, update the existing CLI using the same installation method (rerun the macOS installer or use the returned npm command), within the user's installation authorization. Verify `animaguru --version`, rediscover `animaguru schema`, then retry the original command. `animaguru update` only checks and prints instructions; it does not install. Older CLIs need a manual update once to gain this check. Do not repeatedly retry a blocked command or bypass the check.
+
 For multiple videos with one identity, run `animaguru create workspace NAME`, open that root in Codex or Claude Code, and put logos, fonts and guides in `branding/` or at the root. Configure reusable web assets in `branding/brand.json`. Add a film with `animaguru create animation FILM_NAME` inside the workspace (or `--workspace PATH`). Dependencies, skills, the player and exporters are installed once; each `animations/FILM_NAME/` folder contains only its own film, media and production files. `animaguru create NAME` remains the standalone form for existing workflows.
 
 Before creative work in a multi-animation workspace, select or create the intended animation and read the root branding plus that film's `animaguru.project.json`. Use root `animaguru.templates.json`. If the shared kit is still a placeholder or incomplete, use sibling `animaguru-brand-kit` to obtain the official website or brand guide when the first prompt did not include one, research the identity, and populate `branding/`. Then follow `animaguru-video`: offer **Explainer**, **Advertisement** or **General Video**, then visual style, then the compact brief, reusing supplied answers. A saved brand is shared; purpose, style and story choices belong to each film.
@@ -35,7 +37,7 @@ Generation and reference imports spend credits and require `--yes`. Use it for w
 
 CLI/API balances, estimates, credit ranges and queued-job receipts use AnimaGuru credits, including fractional amounts such as 0.7. This includes offline `--dry-run` estimates. Compare them directly with the user's credit budget; do not rescale returned amounts or convert credits into currency.
 
-`generate`, `jobs submit`, and `files import` accept optional `--image-model` and `--video-model` flags. These select models for that operation and its explicitly queued animations. Use `animaguru models PROJECT_ID --image-model IMAGE_ID --video-model VIDEO_ID` only when the user wants to change defaults for future jobs. Omitted selections use project defaults, then service defaults. Discover available identifiers from `catalog` rather than hardcoding model names.
+Image generation uses staff-managed main and backup routes. Do not choose or send an image model. `generate`, `jobs submit`, and `files import` accept `--video-model` for video selection. Use `animaguru models PROJECT_ID --video-model VIDEO_ID` only when the user wants to change the video default. Discover video identifiers from `catalog`.
 
 ## Standalone images, video, music and speech
 
@@ -44,6 +46,12 @@ CLI 0.4.0 adds `assets generate`, `assets submit`, `assets list`, and `assets do
 ```sh
 animaguru assets generate PROJECT_ID --type image --name "Forest" \
   --prompt "A moonlit watercolor forest" --aspect-ratio 3:2 --yes
+animaguru assets generate PROJECT_ID --type image --name "Marie in the lab" \
+  --prompt "Marie from Image 1, same face, hair and coat, at her lab bench" \
+  --references CHARACTER_SHEET_SHA --aspect-ratio 16:9 --yes
+animaguru assets generate PROJECT_ID --type image --name "Marie, night" \
+  --prompt "Replace the window view with a starry night sky" \
+  --references SCENE_SHA --mask 0.5,0,0.5,1 --yes
 animaguru assets generate PROJECT_ID --type video --name "Forest breeze" \
   --prompt "Leaves drift slowly in the breeze" --reference IMAGE_BLOB_SHA \
   --seconds 5 --resolution 768p --mode quality --yes
@@ -53,11 +61,11 @@ animaguru assets generate PROJECT_ID --type speech --name "Welcome" \
   --prompt "Welcome, traveler." --voice Kore --delivery "Warm and welcoming" --yes
 ```
 
-Use `--model` to select a model for any asset type; discover the choices with `catalog` and the server's configured subset with `assets list`. Asset defaults are per type, independently of sprite project defaults. Image backgrounds are retained unless `--transparent` is supplied. Videos require an image asset's `file` SHA from `assets list`, or a character's `art` SHA from `projects get`, in the same project. Validate video options against the selected model's `videoProfiles` in `catalog`; omitted video settings use five seconds and that provider's first resolution/mode. MiniMax H3 Max and H3 Max Turbo (`fal/minimax-h3-max`, `fal/minimax-h3-max-turbo`) also take `--workflow text-to-video` (with `--aspect-ratio`) and `--workflow extend-video` (with `--source VIDEO_SHA`, 2K resolution, `--extend-output`, and up to three comma-separated `--audio-references`). From images, they accept `--end-reference` with or without `--reference` and a `--soundtrack AUDIO_SHA`. All workflows take `--prompt-expansion` and `--seed`. These need CLI 0.8.0 or newer. For Gemini speech, `--prompt` is spoken verbatim and `--delivery` is a separate instruction. Music uses Lyria; the Gemini TTS models are for speech.
+Images use the staff-managed route automatically. Use `--model` to select a model for video, music, speech or sound effects; discover the choices with `catalog` and the server's configured subset with `assets list`. Asset defaults are per type, independently of sprite project defaults. Image backgrounds are retained unless `--transparent` is supplied. CLI 0.17.0 adds `--references`, `--mask` and the 16:9 and 9:16 image shapes. `--references SHA,SHA` (up to four) makes an image follow this project's saved images, uploads or sprite art. To keep a recurring character consistent, generate a character sheet once and reference its `file` SHA in every later image; refer to references in the prompt as Image 1, Image 2 and so on, and say what to take from each. `--aspect-ratio` is 1:1, 3:2, 2:3, 16:9 or 9:16; use 16:9 or 9:16 for video scenes. `--mask x,y,width,height[;…]` (up to eight rectangles, in fractions of Image 1 from the top left) edits only those parts of Image 1, the first reference, and keeps the rest and Image 1's shape. Look at the image before choosing the rectangles. Videos require an image asset's `file` SHA from `assets list`, or a character's `art` SHA from `projects get`, in the same project. Validate video options against the selected model's `videoProfiles` in `catalog`; omitted video settings use five seconds and that provider's first resolution/mode. MiniMax H3 Max and H3 Max Turbo (`fal/minimax-h3-max`, `fal/minimax-h3-max-turbo`) also take `--workflow text-to-video` (with `--aspect-ratio`) and `--workflow extend-video` (with `--source VIDEO_SHA`, 2K resolution, `--extend-output`, and up to three comma-separated `--audio-references`). From images, they accept `--end-reference` with or without `--reference` and a `--soundtrack AUDIO_SHA`. All workflows take `--prompt-expansion` and `--seed`. These need CLI 0.8.0 or newer. For Gemini speech, `--prompt` is spoken verbatim and `--delivery` is a separate instruction. Music uses Lyria; the Gemini TTS models are for speech.
 
 For ElevenLabs speech, choose `eleven_v4` or `eleven_v4_turbo`, pass an ElevenLabs voice ID with `--voice` (George is the default), and put audio tags in `--prompt` rather than using `--delivery`. For sound effects, use `--type sfx`; the only model is `eleven_text_to_sound_v2`. SFX supports `--seconds` from 0.5 to 30 (default 5), `--loop`, and `--prompt-influence` from 0 to 1 (default 0.3). These additions require CLI 0.5.0 or newer; check `catalog` before using them with an installed CLI.
 
-Use `--dry-run` for local request validation and estimated credits without sign-in or paid work. The estimate does not prove the server has the provider, the credits, or the reference image. Use `--yes` only within the user's existing spending authorization. For file-based requests, write the API asset request as JSON and use `assets submit PROJECT_ID --file asset.json --yes`; `--model` overrides the file's selection.
+Use `--dry-run` for local request validation and estimated credits without sign-in or paid work. The estimate does not prove the server has the provider, the credits, or the reference image. Use `--yes` only within the user's existing spending authorization. For file-based requests, write the API asset request as JSON and use `assets submit PROJECT_ID --file asset.json --yes`; `--model` overrides the file's selection for non-image assets. Image model fields in old files are ignored.
 
 Add `--download PATH` to `assets generate` or `assets submit` to wait for the job and save the result in one call; a folder PATH (existing or ending in `/`) receives `NAME.EXT`, and existing files are never overwritten. `--wait` waits without downloading; `--timeout` sets the limit (default 600 s). For several assets, such as one speech file per voiceover line, write `{"defaults": {"type": "speech", "model": "…", "voice": "…", "delivery": "…"}, "assets": [{"name": "vo-01-hook", "prompt": "…"}]}` and run `assets batch PROJECT_ID --file lines.json --dry-run`, then `--yes --download production/audio/vo/`. Every item is validated before anything is submitted. If a submission fails, the batch stops and prints the receipts already submitted; check those (and `jobs list`) instead of resubmitting. Older CLIs lack these flags: check `--help`, and otherwise use `jobs wait PROJECT_ID JOB_ID`, read the asset id from `assets list` and download with `assets download PROJECT_ID ASSET_ID --output welcome.wav`, or export the complete project ZIP.
 
@@ -71,13 +79,14 @@ animaguru jobs list PROJECT_ID
 animaguru projects get PROJECT_ID
 animaguru files list PROJECT_ID
 animaguru files export PROJECT_ID --output forest-cast.zip
+animaguru files get PROJECT_ID references/ID/document.pdf --output references/lesson.pdf
 ```
 
 `jobs wait` tracks only the specified job. Drawing can queue follow-up animations: inspect `jobs list` and wait for the relevant jobs before exporting. A timeout or CLI exit does not cancel server work. Inspect existing jobs before retrying a paid submission to avoid duplicate charges. Report failed jobs and their errors instead of repeatedly submitting the same work.
 
 `projects move PROJECT_ID --to OTHER_PROJECT_ID [--characters SLUG,SLUG] [--assets ASSET_ID] [--avatars GROUP_ID]` moves work between projects for free: characters with all their animations, saved assets and the project's own avatars. Use it to gather work into one project instead of regenerating it; it needs CLI 0.16.0.
 
-`files export` downloads a ZIP and refuses to overwrite an existing file. Choose a fresh output path. Inspect the exported manifest and actual assets before integrating them into the game or app; use their real dimensions, frame ordering, and animation metadata.
+`files get` downloads one file by its path from `files list`, such as an image, clip or PDF the user attached as a reference. `files export` downloads a ZIP. Both refuse to overwrite an existing file. Choose a fresh output path. Inspect the exported manifest and actual assets before integrating them into the game or app; use their real dimensions, frame ordering, and animation metadata.
 
 ## Shared workspace commands
 
@@ -93,6 +102,10 @@ For hosting, run `animaguru submit animations/NAME --yes`, or `animaguru submit
 
 Pro and Team accounts: `animaguru library search DESCRIPTION [--type sprite|image|video|music|speech|sfx] [--labels a,b]` finds ready-made assets; `animaguru library import ID --project PROJECT_ID` copies one into a project and `animaguru library download ID --output DIR` saves its files. Search it before generating common assets (whooshes, background music, stock clips, characters); importing spends no credits.
 
+## Saved brand kits
+
+`animaguru brands list` lists brands in the selected cloud workspace; `brands get BRAND_ID` reads a kit. Before researching a shared identity, check for the user's saved brand. `brands download BRAND_ID --output WORKSPACE_ROOT` installs its `branding/` files for films; inspect its guide and logo, and use `--force` only when replacing the local kit is intended. Create one with `brands create --name NAME` (or `--file FIELDS.json`), or `brands research OFFICIAL_HTTPS_URL` for a user-supplied website; poll `brands get` until `ready` or `failed`. `brands upload BRAND_ID --file logo.svg --kind logo --yes` uploads an image, then `brands update BRAND_ID --file PATCH.json` can set `{"logo":"bf_…"}` or `{"assets":["bf_…"]}` using returned IDs. Preserve existing asset IDs when extending that list. Pass `--workspace personal` or a team ID to list/create/research for an explicit cloud workspace.
+
 ## On demand
 
 Read these only when the task needs them:

package/skills/animaguru/references/cloud.md

@@ -6,7 +6,7 @@ Read from the `animaguru` skill only when the user asks for cloud sync, versions
 
 CLI 0.11.0 adds `animaguru cloud` for users on Individual Pro or Team. Discover it in `animaguru schema`. Commands run from anywhere inside a video workspace; `--dir PATH` selects another folder.
 
-Only link a folder when the user asks for cloud sync. `cloud link` creates a cloud project in the selected AnimaGuru workspace (`workspaces use` picks a team), and `cloud link --id PROJECT_ID` attaches an existing one. In a linked folder:
+Only link a folder when the user asks for cloud sync. A folder syncs with one AnimaGuru project's animations, shown on that project's Animations tab in the studio. `cloud link --project PROJECT_ID` syncs with an existing project (`prj_…`); without it, `cloud link` creates a new project named after the folder (or `--name`) in the selected AnimaGuru workspace (`workspaces use` picks a team). In a linked folder:
 
 - Check `cloud status` before ending a working session.
 - Save with `cloud push -m "MESSAGE"` at milestones the user would want to return to, such as an approved storyboard, finished scenes or an exported cut. Describe the change in the message.
@@ -14,6 +14,6 @@ Only link a folder when the user asks for cloud sync. `cloud link` creates a clo
 - When push reports a newer cloud version, run `cloud pull`.
 - When pull stops on files changed on both sides, show them to the user. Ask before using `--prefer local` (keep theirs; the next push replaces the cloud copy) or `--prefer cloud` (take the cloud's; local copies go to `.animaguru/sync/backups/`).
 
-`cloud versions` lists history. `cloud restore VERSION [--path P]` brings back an earlier version without rewriting history; then push to save it as the newest version. Restore refuses to replace unsaved changes unless `--yes`, and needs the user's go-ahead. `cloud clone PROJECT_ID DIR` downloads a project; run `npm install` there. `cloud delete PROJECT_ID --yes` permanently removes every version: run it only when the user asks to delete that cloud project.
+`cloud versions` lists history. `cloud restore VERSION [--path P]` brings back an earlier version without rewriting history; then push to save it as the newest version. Restore refuses to replace unsaved changes unless `--yes`, and needs the user's go-ahead. `cloud clone PROJECT_ID DIR` downloads a project; run `npm install` there. `cloud list` lists projects with synced animations. `cloud delete PROJECT_ID --yes` permanently removes the project's animations: every version, Animator draft and stored file, while the project and its assets stay. Run it only when the user asks to delete those animations.
 
 Sync never uploads node_modules, exports, frames, `.env` files, credentials, keys, links or `.gitignore`/`.animaguruignore` matches. Add bulky regenerable files to `.animaguruignore` rather than deleting them. Images, video, audio and ZIP files are stored once in cloud storage at any size and count toward the plan's media storage (20 GB Pro, 100 GB per Team). Every other file is versioned in Git. A work file over 32 MB stops the push: tell the user, and ignore or move it only with their agreement. If the plan ended, pushing is refused, but pulling and cloning still work. Never read or print `.animaguru/sync/` tokens; the CLI requests short-lived Git credentials itself.

package/skills/animaguru-art-motion/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 alchaincyf (花叔 · 花生)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/skills/animaguru-art-motion/SKILL.md

@@ -0,0 +1,35 @@
+---
+name: animaguru-art-motion
+description: Create art-directed AnimaGuru animations using historical and contemporary art materials and explainer motion grammars. Use for animated paintings, art-style journeys, or a requested art-motion, editorial, diagram, whiteboard, presenter, or typographic treatment.
+---
+
+# AnimaGuru Art Motion
+
+Build an original film by choosing how its world is made and how its ideas move. This is one skill with 36 art-style references and nine explainer grammars, adapted into English from Huashu Art Motion. The references are production guidance, not 45 installed renderers or downloadable templates.
+
+## Choose the treatment
+
+Reuse the user's subject, style, brand, reference, language, format and existing authorization. In an AnimaGuru video workspace, follow the sibling `animaguru-video` production contract and reuse its completed brief. An explicit request for this skill is a style choice; do not ask the user to choose it again from the gallery.
+
+- For a material, painting movement or animated art world, open [the art-style index](references/art-styles.md), then only the chosen card.
+- For an explanation, open [the motion-grammar index](references/explainer-grammars.md), then the card matching the passage's explanatory job.
+- For a style journey, keep subject identity and action continuous while changing materials. The same-scene method is in [production](references/production.md).
+- For AnimaGuru's established tactile editorial identity, use sibling `animaguru-cutcraft` when present; [editorial evidence](references/explainers/editorial-evidence.md) supplies document and annotation choreography.
+
+Material and grammar can combine: a mineral-painted world may explain a process through a concept morph; a product film may alternate actual interface footage with a restrained paper diagram. Do not add a grammar that weakens the content simply to display another technique.
+
+## Make the film
+
+Read [production and generation](references/production.md) before producing assets. Pair each narrated idea with a visible event, identify authentic evidence, and establish a strong subject-specific still. Choose procedural drawing, generated layers, generated video or a hybrid according to the work. Use AnimaGuru's actual available models, estimates, asset IDs and jobs; keep essential type and logos exact in the composition.
+
+Read [rendering and review](references/rendering.md) when implementing scenes. Reuse the installed player, scene kit, cue sheet, audio tools and exporter. For PixiJS code, read the available `pixijs` skill first. If the runtime or tools are absent, inspect the project and available capabilities before choosing a local implementation; do not assume upstream `PAINT`, `CAM`, `DG` or `MO` globals are installed.
+
+Motion should serve the selected treatment. An animated painting may have a main action and a few material-specific loops; an evidence passage may hold completely still while a sentence is read. Move the camera to reveal a relation or detail, then let it settle. Repeated slow pushes and repeated fast pushes both become mechanical.
+
+Keep timing seekable, variation seeded and media local. Review the actual film at normal speed, inspect transitions and extreme poses, and verify the export. Report only checks performed. Creation/export does not imply publication or a new paid generation budget.
+
+## Reference use and attribution
+
+The English cards preserve useful construction methods, timing examples, motif motion and failure cases. Source measurements describe upstream experiments; they do not certify a new AnimaGuru implementation. Adapt pixel values to the actual format, English reading pace and brief.
+
+Read [sources and scope](references/sources.md) when extending or redistributing the skill; [coverage.json](references/coverage.json) maps all 45 cards to the pinned source revision. Retain the included [MIT notice](LICENSE). Demo likenesses, third-party mascots, fonts and footage are not included. Use original or authorized assets.

package/skills/animaguru-art-motion/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "AnimaGuru Art Motion"
+  short_description: "Art materials and purposeful explainer motion"
+  default_prompt: "Use $animaguru-art-motion to create an original animated art treatment for my film."

package/skills/animaguru-art-motion/references/art/art-nouveau.md

@@ -0,0 +1,27 @@
+# Art Nouveau
+
+Organic whiplash curves, ornamental halos and integrated botanical decoration.
+
+## Material and construction
+
+Construct variable-width ribbons by offsetting both sides of a sampled centerline. Keep the exterior contour heavier than interior detail. Integrate flowers, hair, steam and framing into a continuous curve language. Use a darker offset silhouette for an outer contour rather than applying equal-width strokes everywhere.
+
+## Palette
+
+Muted botanical greens, cream, warm gold and a restrained rose or rust accent.
+
+## Motion
+
+An ornamental sunrise halo opens, flowers turn, long hair follows a gentle delayed arc and steam traces an S-curve. Attach stems and ribbons to real anchors so they seem to grow from the composition.
+
+## Transition
+
+Grow a botanical gold line into a frame or let an organic ribbon sweep across.
+
+## Failure cases and adaptation
+
+Uniform-width geometry loses the style. Do not cover the explanation with ornament. The source animal silhouette was a weak point: simplify and test it at thumbnail size before adding internal decoration.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/10_nouveau.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/bauhaus-geometry.md

@@ -0,0 +1,27 @@
+# Bauhaus geometry
+
+Primary-color geometry, balanced spatial systems and articulated mechanical movement.
+
+## Material and construction
+
+Compose circles, rectangles, bars and grid modules with clear proportions and pivots. A Schlemmer-inspired geometric figure can share the same primitive vocabulary as the room. Cache the static geometry and retain stable joint locations. Let structure, balance and negative space carry the style.
+
+## Palette
+
+Cream/white, near-black, primary red, blue and yellow.
+
+## Motion
+
+A lever turns, a grid changes colors, a ball contacts a surface and articulated discs rotate around their joints. Use contact timing and mechanical cause-and-effect instead of unrelated floating shapes.
+
+## Transition
+
+A primary-colored geometric iris or moving bar reconfigures the composition.
+
+## Failure cases and adaptation
+
+A generic collection of circles is weak without a strong layout. A head whose pivot drifts disconnects from the body; inspect the neck through all poses. Do not add painterly distress or soft gradients by default.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/12_bauhaus.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/blue-period.md

@@ -0,0 +1,27 @@
+# Blue-period painting
+
+A restrained blue monochrome, elongated figures, broad dry strokes and slow emotional pacing.
+
+## Material and construction
+
+Paint a blue-only base. Use broad low-contrast strokes; source wall values were cell 34, length 110 and width 34 px, followed by a soft pass. Isolate small props instead of excluding rectangular patches from the stroke renderer. Elongate subjects around their feet and use exactly the same matrix for fills, texture and outlines.
+
+## Palette
+
+#132240, #1D3866, #2D5A94, #6B8FB6, #B6C8D8 and teal-gray #3B6878; contours near Prussian blue #0C1830.
+
+## Motion
+
+Use slow breathing, a deliberate hand action, half-speed steam, distant sea bands and moonlight fragments. Source breathing used sin(1.8t) with small amplitude; contours stay steady, without line boiling.
+
+## Transition
+
+Strip away chroma into blue or let a broad blue stroke reveal the scene.
+
+## Failure cases and adaptation
+
+Separate hair, skin and fur through value rather than leftover ochre. Short dense vertical marks look like rain or a blanket; keep strokes broad and quiet. Closely packed round caps can look like rows of heads. Mood and silhouette matter more than a blue color filter.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/36_picasso_blue.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/cave-painting.md

@@ -0,0 +1,27 @@
+# Cave painting
+
+Pigment, charcoal and animal silhouettes on a relief rock surface. Useful for origins, memory and prehistoric worlds.
+
+## Material and construction
+
+Cache a relief-like stone base with broad irregular light and dark patches. Draw charcoal silhouettes and rough ochre deposits over it; let pigment break at the rock grain. Keep the wall texture substantially larger than dust. Use an irregular drawn edge rather than a smooth vector outline.
+
+## Palette
+
+Earth ochre, charcoal, bone and muted red; separate the subject from the rock through value before adding grain.
+
+## Motion
+
+Use a two-pose animal run, fire flicker and rising embers. Let the torch change local illumination without regenerating the wall. A repeated action can continue across style changes using global film time.
+
+## Transition
+
+Reveal pigment along a hand-painted route or let torchlight uncover the next composition.
+
+## Failure cases and adaptation
+
+A modern cartoon face against a convincing cave wall breaks the medium. Simplify human figures into the same pigment language. Fire-driven light variation should not become full-frame random flicker.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/01_cave.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/cinematic-anime-light.md

@@ -0,0 +1,27 @@
+# Cinematic anime light
+
+Detailed environments, luminous summer skies, hard cel shadows and selective optical light.
+
+## Material and construction
+
+Cache architecture, then layer clouds, exterior details, glass, curtains, light patches, subjects and post effects. For a cel shadow, subtract a translated part silhouette from itself; reverse the shift for a narrow rim light. Cloud groups need a blue-gray common silhouette plus offset lit lobes, then soft/clear passes, rather than outlining every sphere.
+
+## Palette
+
+Warm skin #FFE8DC / #F2B6A8, cool white #FBFBFE / #B9C3DC, navy #2C3A66, saturated sky and warm sunlight.
+
+## Motion
+
+Clouds slowly develop, a contrail and birds cross, a glass droplet slides with a trace, and a curtain’s shadow follows its fabric. Let hair and leaves share the breeze. Cache blurred light wedges; use sparse dust and restrained quarter-resolution bloom.
+
+## Transition
+
+A diagonal light wipe or match cut along the bright sky.
+
+## Failure cases and adaptation
+
+Keep the sun inside the glass opening so flare does not appear to originate on the wall. Oversized horizontal flares resemble stray lines. Draw a droplet with transparency, a lower arc and a glint rather than an opaque bubble. The source had strong lighting but limited scene/character detail.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/35_shinkai.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/constructivist-poster.md

@@ -0,0 +1,27 @@
+# Constructivist poster collage
+
+Cream paper, red wedges, black geometry, photomontage and forceful type.
+
+## Material and construction
+
+Convert photos to gray halftone on a 45-degree grid. Radius can follow cell × gain × sqrt(1−luminance), batching dots into one path. Source cell/gain values were 4.6/.74 for faces and 7/.8 for background. Give cutouts a clear paper edge, then compose diagonals, circles and modular lettering.
+
+## Palette
+
+Cream #E9DFC4, red #C8231C / #8F1711, black #151311 and light paper #F4ECD6.
+
+## Motion
+
+Stamp words in with a short stagger, animate a red wedge as a speech or broadcast cone, and let slogans wave after landing. Keep a radio arc or one moving geometric element subordinate to the message.
+
+## Transition
+
+A diagonal red wedge, typographic bar or printed stamp changes the scene.
+
+## Failure cases and adaptation
+
+Oppositely wound letter components may cancel holes under nonzero fill; fill components separately where needed. Accents must fit the line box. Use glyphs with actual font coverage. Photo halftone is evidence texture only when the underlying material is authentic.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/22_constructivism.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/contemporary-flat.md

@@ -0,0 +1,27 @@
+# Contemporary textured flat illustration
+
+Clean silhouettes, regional stipple and expressive timing.
+
+## Material and construction
+
+Compose bold flat regions with little or no outline. Add seeded stipple whose density varies within a shape, then cache it. Build expressive pose targets appropriate to the subject; use generated character frames when detailed acting matters. Keep the same material treatment on subject and environment.
+
+## Palette
+
+A small contemporary editorial palette with clear value separation; use the client brand when relevant.
+
+## Motion
+
+Choose an actual performance such as a jump, reach or surprised recoil. Use anticipation, contact, compression and delayed secondary follow-through. Anchor squash and stretch at the feet or other support point.
+
+## Transition
+
+A concentric iris or bold shape reveal echoes the composition.
+
+## Failure cases and adaptation
+
+Do not stretch around the center and make feet slide through the ground. Large poses may need their own silhouettes rather than forcing a single rig. Stipple is surface material, not a reason to regenerate noise each frame.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/16_2026.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/cosmic-comic.md

@@ -0,0 +1,27 @@
+# Cosmic four-color comic
+
+Offset color plates, spotted blacks, feathered hatching, energy and exaggerated perspective.
+
+## Material and construction
+
+Draw color and ink separately. Halftone color plates can be offset about (4,3) px; composite a white-filled, black-ink pass with multiply. Make a dark crescent by subtracting a translated part from itself, then add a wider clipped hatching band. Source contours were about 6 px, hatching about 2.6 px at 9 px spacing.
+
+## Palette
+
+Cream paper #F4ECD6, CMYK-like flats, strong black and energetic yellow/orange.
+
+## Motion
+
+Animate clusters of black energy circles: denser and larger at the center, changing at about 12 fps. Let foreshortened hands or a prop lead speed lines, and reserve extruded onomatopoeia for an impact. A reaction inset can follow the main action.
+
+## Transition
+
+A comic panel shift, energy burst or angular impact wipe.
+
+## Failure cases and adaptation
+
+Letter bursts can cover faces. Split a planet ring into back and front arcs. The upstream sample remained closer to clean pop art: more forceful foreshortening, block-shaped fingers and muscular tension are needed for this specific reference language.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/27_kirby.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/cubist-facets.md

@@ -0,0 +1,27 @@
+# Cubist facets
+
+Multiple viewpoints, angular silhouettes and overlapping planes.
+
+## Material and construction
+
+Start with structural facets large enough to explain the form. Layer translucent gradient planes; let selected facets extend past the original contour to change the silhouette. Use a coarse free-edge mesh rather than clipping a fine triangular overlay to an otherwise realistic object. A face can combine a profile and frontal plane intentionally.
+
+## Palette
+
+Muted ochre, brown, blue-gray and cream; keep a deliberate focal value structure.
+
+## Motion
+
+Shuffle letter fragments, slide window planes and let a paper facet tremble at a pinned edge. Animate a few structural planes, retaining recognizability through the change.
+
+## Transition
+
+Break or reassemble the picture into angular shards.
+
+## Failure cases and adaptation
+
+A low-poly filter alone does not express multiple viewpoints. Too many bright brown masses flatten the subject, an issue in the upstream demo. Keep facial landmarks and large planes readable, with no continuously random triangle reshuffle.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/11_cubism.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/dunhuang-mural.md

@@ -0,0 +1,27 @@
+# Dunhuang-inspired mineral mural
+
+Mineral pigment, iron-wire contour, flying ribbons and aged plaster.
+
+## Material and construction
+
+Lay mineral colors over plaster with thin warm-dark contours, around 2.6 px in the source scale. Age the background with broad peeling masks, protecting faces and garments. Use translucent halos and flowing overhead ribbon arcs. Inspect historical reference for costume and iconography rather than inventing factual details.
+
+## Palette
+
+Mineral red #A5452F / #7A3020, green #3F8F6E / #6AB08E, blue #2F5F8F / #5A86B0, white #ECDFC4 and gold #D8B05A.
+
+## Motion
+
+A long ribbon travels through the scene, flowers rotate and fall, lamps flicker and a few plaster flakes descend. Give ribbons enough travel to be visibly flowing; an 18 px sway was ineffective compared with a much broader arc in the source.
+
+## Transition
+
+A flying ribbon sweeps across and becomes the next composition.
+
+## Failure cases and adaptation
+
+Fine random speckles look like dirt rather than broad plaster loss. Ageing on a body reads as damaged clothing. An opaque ring looks like a life buoy rather than a halo. Character anatomy remained a weaker part of the upstream demo.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/20_dunhuang.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/early-cgi.md

@@ -0,0 +1,27 @@
+# Early CGI and ray-traced plastic
+
+Glossy primitives, chrome, checkerboard reflections and conspicuously synthetic surfaces.
+
+## Material and construction
+
+Use a coherent plastic shading model: base color, radial dark edge, environment reflection and hard highlight. Cylindrical limbs and rounded solids should share the same light source. A mirrored floor image can suggest reflection in a 2.5D scene; describe it as a controlled approximation. Use a real 3D renderer when actual ray-traced geometry is required.
+
+## Palette
+
+Bright plastic colors, chrome gray, a dark/light checkerboard and sharp white highlights.
+
+## Motion
+
+Rotate a torus, turn a reflective object and animate a glint or mechanical eye. Reflection motion should follow the subject rather than behave as an independent overlay.
+
+## Transition
+
+Wireframe resolves into shaded geometry, or a scanline reveals the next scene.
+
+## Failure cases and adaptation
+
+Blocky faces and legs weakened the source demo. Avoid claiming physically correct ray tracing for a painted highlight. Keep reflections aligned to the floor and contact points; a displaced mirror image breaks depth immediately.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/15_raytrace.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/egyptian-wall-painting.md

@@ -0,0 +1,27 @@
+# Egyptian wall painting
+
+Registers, profile silhouettes, mineral pigment and deliberate symbolic movement.
+
+## Material and construction
+
+Build a warm plaster ground, then horizontal registers, architectural borders and flat figures. Combine a profile head with a frontal eye and shoulders deliberately. Use fine consistent contours and repeating glyph bands; draw or source accurate symbols when their meaning matters.
+
+## Palette
+
+Sand and cream ground with mineral red, blue, green, black and restrained gold.
+
+## Motion
+
+A scarab pushes a sun disc; rays open like hands; a glyph band rewrites or scrolls. Articulated arms should turn at clear joints. Quantize decorative actions if a ceremonial stepped rhythm helps.
+
+## Transition
+
+A painted fabric band or vertical register wipes across the scene.
+
+## Failure cases and adaptation
+
+Do not treat decorative glyphs as a translated quotation. Stiff elbows can be improved with measured shoulder, elbow and wrist anchors. Keep the intentional frontal/profile convention consistent instead of drifting between views.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/02_egypt.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/expressionist-waves.md

@@ -0,0 +1,27 @@
+# Expressionist waves
+
+Long translucent currents, pressure in the sky and controlled whole-scene distortion.
+
+## Material and construction
+
+Integrate long flow-field strokes of roughly 200–300 px rather than short outlined impasto. Source examples use 26–42 steps, 6–8 px per step, 9–13 px brush width and alpha .5–.78. Keep the flat color below translucent strokes. Apply continuous row and column warps with overscan at least as large as the maximum displacement.
+
+## Palette
+
+Burning sky oranges/reds such as #E2541C, #F08A24, #C8301E and #B82A1E against cool water and dark structural masses.
+
+## Motion
+
+Let the sky undulate and the environment bend around the focal action. One source warp combines 9 sin(.012y+4.2t) and 4 sin(.031y−6t), with a smaller column warp. Scale all amplitudes to the output and desired emotional intensity.
+
+## Transition
+
+A downward pressure of color or a wave carries the cut.
+
+## Failure cases and adaptation
+
+Short dark-rimmed strokes make this look like the post-impressionist card. Thin translucent currents are the distinction. Overscan prevents exposed blank strips. Choose type that supports the mood rather than a comic handwritten default.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/19_munch.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/flat-outline-cartoon.md

@@ -0,0 +1,27 @@
+# Flat outline television cartoon
+
+A Rick-and-Morty-inspired drawing treatment using original characters and semantic parts.
+
+## Material and construction
+
+Use hard flat fills, sparse same-hue shadows, rounded contours and low-contrast background mottling. Source contours were #222323 at 2.35 px, with details 1.6–2.6 px at 1920 width. Separate body, head, arms/prop, eyelids and pupils into semantic groups. Establish a convincing still, then animate existing anchors. Build a tail from a centerline and width profile.
+
+## Palette
+
+Warm gray wall #CDBF96, wood #9B7752, skin #EFC4A0 and sky #A7D6EA are source examples; adapt them to the new subject.
+
+## Motion
+
+Move gaze, deterministic blinks, a small head turn, a slow prop lift and a continuous tail. Keep the shoulder anchored while the sleeve follows the hand. The source jump moves body and tail together; it contains no full walk cycle or landing compression.
+
+## Transition
+
+An intentional cut, prop match or simple scene-camera change; traveling through other art styles is optional.
+
+## Failure cases and adaptation
+
+The upstream example is standalone SVG, unlike the 35 Canvas scenes. Do not assume its renderer API exists in AnimaGuru. Rebuild semantic groups in the current runtime or rasterize intentionally. SVG DOM state can be deterministic while screenshots vary with partial rasterization; verify the actual capture configuration. Do not reuse show characters, voices or upstream demo likenesses.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/37_rick_morty.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).

package/skills/animaguru-art-motion/references/art/gothic-illumination.md

@@ -0,0 +1,27 @@
+# Gothic illuminated manuscript
+
+Vellum, embossed gold leaf, stained glass, ornamental borders and restrained marginal humor.
+
+## Material and construction
+
+Cache vellum, architecture, gold, borders and text columns. Make gold with low-frequency variation plus strongly horizontal polishing grain. A 46 px diamond grid with dark lines and highlights offset about 1.2 px suggests embossing. Build a gold-only mask from the painted regions or a material ID; sweep a diagonal light through that mask. Keep colored parchment out of the mask.
+
+## Palette
+
+Gold #9C6A1A / #D9A63A / #F8DE86, warm vellum, manuscript red and blue.
+
+## Motion
+
+Use a steady gold sweep, glints when it crosses raised decoration, swaying vines, stained-glass sparkle and a small marginal creature. A constant-speed sweep can place the glint more reliably than an easing curve that races past the subject.
+
+## Transition
+
+Turn a manuscript page or sweep gilding into the next scene.
+
+## Failure cases and adaptation
+
+Measure the actual text width; blackletter easily overflows narrow columns. Leaves need enough scale and density to read at delivery size. Keep gold sparkle localized, and measure local motion after any camera entrance settles.
+
+Parameters above are upstream study values at approximately 1920×1080 unless stated otherwise. Scale them to the delivery format and check the actual result. These are English production adaptations, not bundled renderers or claims of validated AnimaGuru implementations.
+
+Adapted from [the upstream card](https://github.com/alchaincyf/huashu-art-motion/blob/2571ee0a9a0467c4d99e935ea9e0fadda19210b0/references/%E9%A3%8E%E6%A0%BC%E9%85%8D%E6%96%B9/05_gothic.md) under the [MIT license](../../LICENSE). See [provenance](../sources.md).