@autoaiai/aicut 0.1.11 → 0.1.13

package/skills/personal-ip-brand-intro/SKILL.md

@@ -1,216 +0,0 @@
----
-name: personal-ip-brand-intro
-description: "Create original commercial-quality personal-IP brand intro videos with typography, original vector illustrations, optional user-provided images, and either uploaded-music synchronization or a self-designed silent beat grid. Use for 16:9, 9:16, or 1:1 creator openers, kinetic identity films, animated brand title sequences, and AICUT or AICUT production that must be designed for each brand rather than filled into a fixed template. Runs in AICUT: the film is film.html, a page per shot, drawn by `aicut look` and `aicut render`."
-license: MIT
-source: "toufuim/personal-ip-brand-intro-skill"
----
-
-# Personal IP Brand Intro
-
-## Running it in AICUT
-
-AICUT's film is `film.html`: it cuts pages, footage, pictures and sound together on tracks. A page is a web page that
-sets `window.film = { frame(t) }` and draws that moment — no framework, no build step, no composition object, no
-`registerRoot`. Five shots are five pages cut together, not one page that changes scene, so each shot stays trimmable
-and replaceable in Studio.
-
-```sh
-npx -y @autoaiai/aicut@latest          # the manual: the format and the commands, always current
-npx -y @autoaiai/aicut@latest open .   # make this folder a film and print Studio's address
-npx -y @autoaiai/aicut@latest look     # a contact sheet of the whole film, its sound, and a determinism check
-npx -y @autoaiai/aicut@latest look 4.5 # one frame, full size — look at the PNG, not at the code
-npx -y @autoaiai/aicut@latest render   # the MP4, beside film.html
-```
-
-Run the manual before the first film: it is authoritative for the format, and this skill is about the method above it.
-
-Where the words below name another program's part, it becomes this:
-
-| elsewhere | in AICUT |
-|---|---|
-| composition, timeline, project file | `film.html`; tracks and clips inside it |
-| a scene, a sequence, a slide | a page (`pages/<name>.html`) cut onto a track |
-| `useCurrentFrame()`, `interpolate()` | `frame(t)` — the picture depends only on `t`, never on state or `Date.now()` |
-| a still, a screengrab | `aicut look <t>` |
-| `render` / `build` / `export` | `aicut render` (the person can also export from Studio) |
-| a dev server, a preview window | `aicut open` and the Studio address it prints |
-
-Hard rules that hold whatever the method says: the picture depends only on `t`; every value in the person's prompt is
-binding; look at the frames `look` writes before saying a film is done.
-
-
-Build a short identity film in which typography, optional images, original illustrations, graphics, and cuts express one creator's brand. Treat every request as a new design system. Use demonstration projects only to validate the workflow, never as templates.
-
-## Hard Rules
-
-- Do not require a logo, portrait, photo, or character image. Support text-and-illustration-only work.
-- Use an uploaded image only when the user provides it or explicitly approves generation.
-- Preserve every uploaded source unchanged; create separate working copies.
-- Use an uploaded track when provided. When none is provided, design a silent beat grid instead of searching for or silently substituting music.
-- Keep every essential brand message readable during a clean hold.
-- Do not reuse demonstration-brand copy, palettes, layouts, scene orders, timing maps, portraits, logos, or motifs.
-- Do not publish, upload, overwrite, or formally render without the required user authorization.
-
-## Route the Work
-
-- Read `references/intake-schema.md` before writing the brief or `brand-spec.json`.
-- Read `references/rhythm-routing.md` to choose uploaded-music or designed-beat-grid timing.
-- Read `references/music-upload-workflow.md` only when the user provides audio or a video containing audio.
-- Use `$music-to-video` only for uploaded-track analysis.
-- Use `$hyperframes` as the entry point for HyperFrames production, then load its creative, animation, and CLI skills as needed.
-- Use Remotion for an explicitly requested React composition or dual-engine delivery.
-- Use `$imagegen` only when the user asks for a new visual asset or accepts generation as a proposed fallback.
-
-## Workflow
-
-### 1. Collect Identity-Critical Inputs
-
-Require:
-
-- brand name
-- creator name, handle, or public role
-- creator category
-- two to four services or content pillars
-- tagline or permission to develop one
-- aspect ratio and desired duration
-
-Treat images, logos, and music as optional. Ask only for missing identity-critical information. Infer reversible creative details when the user requests autonomous production. Never invent certifications, results, clients, credentials, or product claims.
-
-### 2. Select the Visual Mode
-
-Record one mode in `brand-spec.json`:
-
-- `text-illustration`: kinetic typography plus original SVG/CSS/vector illustration; no external image required
-- `image-assisted`: one or more user-provided portraits, character images, product images, or logos carry identity
-- `mixed`: typography, original illustration, and user-provided images share the system
-
-For `text-illustration`, invent category-native visual metaphors rather than generic decorative stickers. For image modes, make the supplied media part of the identity story rather than a corner decoration.
-
-### 3. Select the Timing Mode
-
-Choose exactly one route:
-
-#### Uploaded music
-
-Preserve the upload in `assets/source/`, create a separate working track, and follow `references/music-upload-workflow.md`. Analyze the final working track with `$music-to-video`. Treat its `audiomap.json` as the only timing authority.
-
-#### No uploaded music
-
-Choose a BPM and accent pattern from brand personality and communication density. Create the canonical silent beat grid:
-
-```bash
---duration 7 \
-  --bpm 116 \
-  --output audiomap.json
-```
-```
-Record `music.mode` as `none` and `timing.mode` as `designed-beat-grid`. Do not add, generate, search for, or recommend music unless the user separately asks for it. A silent video can still have precise visual rhythm.
-```
-
-### 4. Establish the Production Mode
-
-Default to showing one recommended visual direction and a concise storyboard before animation and formal rendering. If the user explicitly asks for autonomous execution, post the direction as a progress update and continue.
-
-Support `hyperframes`, `remotion`, or both. Use HyperFrames when the user does not choose. Build both only when dual-engine delivery is explicitly requested.
-
-### 5. Normalize the Brand Brief
-
-Create `brand-spec.json` from the matching example:
-
-- `assets/brand-spec.example.json` for text-and-illustration with no music
-- `assets/brand-spec.uploaded.example.json` for user images and uploaded music
-
-Validate:
-
-### 6. Map Rhythm to Meaning
-
-Use the canonical `audiomap.json`, whether analyzed or designed, to map:
-
-- hook: interrupt attention
-- identity: reveal creator, category, or brand
-- proof: show services, process, tools, or knowledge
-- point of view: land the tagline or belief
-- ownership: resolve the final brand lockup
-
-Do not cut on every beat. Use strong accents for hierarchy changes, weaker accents for secondary motion, and deliberate holds for reading.
-
-### 7. Create an Original Visual System
-
-Read `references/originality-rules.md`. Define a one-sentence visual thesis and a seven-part fingerprint:
-
-- composition grammar
-- type personality
-- graphic or illustration vocabulary
-- image/subject treatment
-- color logic
-- transition family
-- rhythm behavior
-
-Generate three internal directions: category-native, personality-native, and audience-native. Build the option with the clearest brand recognition, strongest timing fit, and lowest similarity to prior work.
-
-### 8. Write the Beat-Synced Storyboard
-
-Read `references/storyboard-contract.md`. For each scene record:
-
-- exact time range from `audiomap.json`
-- communication goal
-- exact on-screen copy
-- text, image, or illustration role
-- dominant motion
-- rhythmic anchor
-- exit or handoff
-
-Keep essential text complete. Cropping may occur only during an intentional entrance or exit. Give the final lockup a clean readable hold of roughly 0.8–1.2 seconds.
-
-### 9. Build the Composition
-
-Create an isolated project and preserve all original media.
-
-For HyperFrames:
-
-1. initialize or resume a clean project
-2. derive all timing from the canonical map
-3. implement seek-safe, frame-derived animation
-4. run `hyperframes check` before rendering
-
-For Remotion:
-
-1. register one explicit composition per intro
-2. derive timing from `useCurrentFrame()` and composition FPS
-3. attach `<Audio>` only for approved uploaded music
-4. keep copy, duration, timing anchors, and creative thesis aligned with the brand spec
-
-When building both engines, share the brand spec, timing map, storyboard, copy, duration, and creative thesis. Do not mechanically copy engine-specific implementation details.
-
-### 10. Verify and Render
-
-Read `references/qa-checklist.md`. Validate layout, motion, contrast, exact text, optional image treatment, rhythm, final metadata, representative frames, and originality.
-
-Verify each final MP4:
-
-```bash
-/absolute/path/to/output.mp4 \
-  /absolute/path/to/brand-spec.json
-```
-```
-Inspect frames extracted from the rendered MP4, not only browser previews. Render only after blocking checks pass and the user has authorized formal rendering.
-```
-
-## Deliver
-
-Return:
-
-- final MP4 or dual-engine outputs
-- `brand-spec.json`
-- canonical `audiomap.json`
-- beat-synced storyboard
-- source project
-- representative frames or contact sheet
-- concise QA summary
-
-For uploaded music, identify the exact user source and selected time range without redistributing the source separately. For designed beat grids, state the chosen BPM, accent logic, and that the output is silent.
-
-## Source
-
-Rewritten for AICUT from [`toufuim/personal-ip-brand-intro-skill`](https://github.com/toufuim/personal-ip-brand-intro-skill)
-(MIT), personal-ip-brand-intro/SKILL.md. The method is that skill's own; the way it runs is AICUT's.

package/skills/personal-ip-brand-intro/references/intake-schema.md

@@ -1,79 +0,0 @@
-# Intake and Brand Spec
-
-Use this schema to turn a conversational request into a production brief.
-
-## Required
-
-| Field | Type | Rule |
-|---|---|---|
-| `brand` | string | Exact public brand name |
-| `creator` | string | Creator name, handle, or public role |
-| `category` | string | Specific creator category |
-| `services` | string[] | Two to four offers or content pillars |
-| `tagline` | string | One clear point of view |
-| `style` | string | Descriptive original visual thesis |
-| `palette` | string[] | Three to six hex colors |
-| `rhythm` | string | Semantic rhythm in four or more stages |
-| `aspect` | string | `16:9`, `9:16`, or `1:1` |
-| `duration` | number | Seconds; normally 5–12 |
-| `engine` | string[] | One or both of `hyperframes`, `remotion` |
-| `visualMode` | string | `text-illustration`, `image-assisted`, or `mixed` |
-| `media` | object | Optional image assets; `images` may be empty |
-| `music` | object | `uploaded` or `none` provenance |
-| `timing` | object | `uploaded-music` or `designed-beat-grid` |
-
-## Text-and-Illustration Example
-
-```json
-{
-  "visualMode": "text-illustration",
-  "media": {"images": []},
-  "music": {
-    "mode": "none",
-    "source": null,
-    "working": null,
-    "userUploaded": false
-  },
-  "timing": {
-    "mode": "designed-beat-grid",
-    "bpm": 116,
-    "beatsPerBar": 4,
-    "offset": 0,
-    "accentPattern": ["strong", "weak", "medium", "weak"],
-    "map": "audiomap.json"
-  }
-}
-```
-
-## Image-and-Music Example
-
-```json
-{
-  "visualMode": "image-assisted",
-  "media": {
-    "images": [
-      {
-        "role": "portrait",
-        "source": "/absolute/path/to/user-portrait.png",
-        "working": "assets/portrait.png",
-        "origin": "user-provided"
-      }
-    ]
-  },
-  "music": {
-    "mode": "uploaded",
-    "source": "/absolute/path/to/user-upload.mp3",
-    "working": "assets/bgm.mp3",
-    "userUploaded": true,
-    "segment": {"start": 0, "duration": 7}
-  },
-  "timing": {
-    "mode": "uploaded-music",
-    "map": "audiomap.json"
-  }
-}
-```
-
-## Missing Information
-
-Ask for missing identity-critical copy. Images and music are optional. If images are absent, use `text-illustration`. If music is absent, use `designed-beat-grid`. Infer only reversible creative details and label them as creative choices.

package/skills/personal-ip-brand-intro/references/music-upload-workflow.md

@@ -1,40 +0,0 @@
-# User-Uploaded Music Workflow
-
-Read this file only when the user provides audio or a video containing audio.
-
-## Source Policy
-
-The user owns the music choice. Accept the provided media and never replace it with searched, stock, demonstration, remembered, or generated music.
-
-## File Layout
-
-```text
-assets/
-  source/
-    user-upload.<original-extension>
-  bgm.mp3
-audiomap-source.json
-audiomap.json
-```
-
-Preserve `assets/source/user-upload.*` byte-for-byte. Never overwrite it.
-
-## Duration Decisions
-
-1. If the source duration matches the requested intro within 0.15 seconds, use the full source.
-2. If longer, analyze the full source and select a coherent phrase with a clear entry and resolution.
-3. Record the selected source start and duration in `brand-spec.json`.
-4. Create `assets/bgm.mp3`, analyze it, and use only its `audiomap.json` for animation.
-5. If shorter, ask whether to shorten the video or upload another track.
-
-Do not loop, stretch, or pitch-shift music without explicit approval.
-
-## Editing Rules
-
-- Major section change: change hierarchy, scene, or dominant motif.
-- Medium onset: reveal a service, illustration, image treatment, or secondary word.
-- Weak beat: micro-motion only.
-- Silence or decay: hold, simplify, or resolve.
-- Final hit: settle the brand lockup rather than start new information.
-
-Report the chosen time range when a long source was trimmed. Do not redistribute the user's source as a standalone deliverable.

package/skills/personal-ip-brand-intro/references/originality-rules.md

@@ -1,42 +0,0 @@
-# Originality Rules
-
-The workflow is reusable; the visual result is not a reusable template.
-
-## Brand Fingerprint
-
-Before storyboarding, define all seven dimensions:
-
-1. Composition grammar: split screen, editorial spread, modular grid, cinematic stage, data HUD, collage, or another system.
-2. Type personality: serif authority, grotesk energy, technical mono, handwritten warmth, or a justified combination.
-3. Graphic vocabulary: category-specific objects, diagrams, textures, strokes, cards, icons, or illustration forms.
-4. Image/subject treatment: define how supplied media participates, or explicitly define the typography-to-illustration relationship for text-only work.
-5. Color logic: background/foreground roles and one deliberate accent behavior.
-6. Transition family: wipes, masks, page turns, hard cuts, shutters, folds, scans, or another coherent family.
-7. Rhythm behavior: impact-heavy, progressive reveal, elastic, percussive, restrained, or another music-linked cadence.
-
-## Anti-Template Test
-
-A new brand must differ from any example on at least five of seven fingerprint dimensions. Merely changing copy, portrait, colors, or icons does not pass.
-
-Reject a direction when:
-
-- the same scene order could accept another brand with only text replacement
-- the dominant motif has no connection to the category
-- all beats use the same entrance animation
-- supplied media is decorative rather than part of the identity story
-- copy is obscured to simulate energy
-- the final lockup resembles a demonstration brand
-
-## Controlled Reuse
-
-Reusable elements may include validation scripts, data schemas, safe-area helpers, audio-map parsing, rendering commands, and generic animation utilities. Do not bundle or reuse demonstration portraits, copy, logos, palettes, compositions, timing maps, or visual assets.
-
-## Three-Direction Exercise
-
-Generate three short internal directions:
-
-- one category-native direction
-- one personality-native direction
-- one audience-native direction
-
-Select the direction with the clearest brand recognition, strongest music fit, and lowest similarity to prior work.

package/skills/personal-ip-brand-intro/references/qa-checklist.md

@@ -1,60 +0,0 @@
-# QA Checklist
-
-## Automated
-
-- Brand spec validator passes.
-- HyperFrames layout, motion, and contrast checks have zero errors.
-- Remotion composition resolves with the intended width, height, FPS, and duration.
-- Final MP4 contains H.264 video.
-- Uploaded-music mode contains AAC audio; designed-beat-grid mode may be silent.
-- Duration is within 0.15 seconds of the brand spec.
-- Aspect ratio and orientation match the request.
-
-## Visual
-
-Inspect rendered-MP4 frames at the hook, identity, proof, point of view, and final lockup.
-
-- No missing, clipped, substituted, or garbled essential text.
-- Supplied images, when present, remain recognizable and are not unintentionally distorted.
-- Text-only work contains a meaningful category-native illustration system.
-- The largest text has sufficient breathing room.
-- Every major scene has a clear hierarchy.
-- Decorative graphics support rather than compete with the message.
-- Transitions have clean before/after states.
-- Final lockup is readable without pausing.
-
-## Audio and Rhythm
-
-For uploaded music:
-
-- Confirm `music.mode` is `uploaded` and `music.userUploaded` is `true`.
-- Confirm source and working track are separate files.
-- Confirm audio is present, correctly trimmed, and not unexpectedly silent.
-- Confirm the reported source time range matches `brand-spec.json`.
-
-For no uploaded music:
-
-- Confirm `music.mode` is `none`.
-- Confirm `timing.mode` is `designed-beat-grid`.
-- Confirm the MP4 is silent unless the user separately authorized audio generation.
-- Report the chosen BPM and accent cycle.
-
-For both:
-
-- Major visual changes land on mapped events.
-- Not every element moves at once.
-- Holds exist between impacts.
-- The ending resolves rather than cutting accidentally.
-
-## Originality
-
-- Compare the seven fingerprint dimensions in `originality-rules.md`.
-- Confirm at least five dimensions differ from every demonstration or recent brand.
-- Confirm no demonstration copy, palette, layout, timing map, image, or visual asset is present.
-
-## Delivery
-
-- Include the final MP4, source project, brand spec, timing map, and proof frames.
-- State checks performed and accepted warnings.
-- Do not call a preview or lint result a finished render.
-- Do not redistribute user-provided music or images separately from the user's project.

package/skills/personal-ip-brand-intro/references/rhythm-routing.md

@@ -1,41 +0,0 @@
-# Rhythm Routing
-
-Choose one timing authority and keep it canonical for the entire project.
-
-## Uploaded Music
-
-Use when the user supplies audio or a video containing audio.
-
-- Preserve the source.
-- Create a working copy.
-- Analyze the final working segment.
-- Use the resulting `audiomap.json` for every cut and motion anchor.
-- Do not use a second analyzer or hand-adjust the map without recording the change.
-
-## Designed Beat Grid
-
-Use when no music is supplied.
-
-Choose a tempo by brand behavior:
-
-| Character | Starting range |
-|---|---|
-| calm, premium, reflective | 82–100 BPM |
-| educational, editorial, conversational | 100–118 BPM |
-| energetic, tech, youth-oriented | 118–136 BPM |
-
-These are starting ranges, not genre labels. Adjust for text density and duration. Prefer 4/4 unless another meter clearly supports the concept.
-
-Create the timing map with `scripts/create-beat-map.mjs`. The default accent cycle is `strong,weak,medium,weak`. Use strong beats for scene or hierarchy changes, medium beats for proof points, weak beats for micro-motion, and bar endings for holds or handoffs.
-
-The video remains silent. Do not create an audio file merely to justify the beat grid. If the user later requests music, treat that as a new uploaded or explicitly generated-audio decision and rebuild the canonical timing map.
-
-## Semantic Timing
-
-Whether timing comes from audio or design:
-
-- reserve the first 0.5–1.0 seconds for the hook
-- leave readable holds after major text entrances
-- cluster related service reveals
-- simplify before the final lockup
-- give the final identity a stable 0.8–1.2 second hold

package/skills/personal-ip-brand-intro/references/storyboard-contract.md

@@ -1,42 +0,0 @@
-# Storyboard Contract
-
-## Typical 7-Second Arc
-
-This is a communication arc, not a fixed scene template.
-
-| Beat | Purpose | Typical duration |
-|---|---|---|
-| Hook | Pattern interrupt or provocative statement | 0.5–1.0 s |
-| Identity | Creator, category, or brand recognition | 0.8–1.4 s |
-| Proof | Offers, tools, process, or expertise | 1.0–1.8 s |
-| Point of view | Memorable belief or tagline | 0.8–1.5 s |
-| Lockup | Brand ownership and recall | 0.8–1.2 s |
-
-Add, merge, or reorder beats when the music or brand calls for it.
-
-## Scene Record
-
-```json
-{
-  "id": "S03",
-  "time": [2.1, 3.35],
-  "goal": "Show three content pillars without becoming a list slide",
-  "copy": ["內容策略", "案例拆解", "實作工作流"],
-  "visualRole": "Typography anchors the frame while an original diagram explains the process",
-  "motion": "Three distinct beat-driven reveals, then a short hold",
-  "audioCue": "Hits at 2.18, 2.61, and 3.04 seconds",
-  "exit": "Diagram line becomes the next scene's underline"
-}
-```
-
-## Text Safety
-
-- Use exact approved wording.
-- Keep essential text inside a 5% safe area unless the design deliberately uses edge typography.
-- A mask may hide text while entering or exiting, never during its required reading hold.
-- Do not split Traditional Chinese words arbitrarily.
-- Check the rendered MP4, not only the source or preview.
-
-## Visual Role
-
-Typography, illustrations, and optional images should explain category, process, proof, or personality. Decorative elements must reinforce the same visual language. Prefer two or three recurring motifs over a random assortment of stickers.