agentleFS
alisadiq-aialisadiq-ai/favstash-ai-video-editing-skills/skills/3-styles-animated/film-styles/SKILL.md

film-styles

Skill75 starsChanged 40 days agoApache-2.0View on GitHub

Style for edit-video: 99 film looks your agent draws in code, each as a 9:16 reel or a 16:9 YouTube explainer. Whiteboard, stick-figure and tablet-sketch explainers, chalkboard, watercolour, ink wash, 80s cel anime, rubber-hose cartoons, manga panels, pixel RPG, newsreel, caption news, data stories, 3D map explainers, isometric infographics, cutaways, tilt-shift miniatures, brick toys, risograph, neon tubes and more, grouped by what the picture looks like. Pick one, bring a story, and HyperFrames renders it on your machine: no camera, no video model, no stock footage.

The file

Files
SKILL.md

Film styles

A library of 99 finished looks. Each style is one STYLE.md: what the picture is made of, its colour logic, type, motion, camera, sound, the moves only that medium can do, its pitfalls, and its two formats:

  • Reel (1080×1920): how the look fills a phone screen, what is on frame 1 and what makes people stay by three seconds.
  • YouTube (1920×1080): the landscape layout, how the look carries a 2–8 minute explainer in chapters, the first ten seconds and the thumbnail frame.

Your agent reads one, writes a treatment, and draws every frame in code. edit-video owns the voice, the sound mix and the review; this skill owns the look.

Pick a style

The style index lists all 99 by group. Name a style by its name or its folder, and say the format if it isn't a reel:

Make a 30-second reel in the Whiteboard style about how a credit score works.

Make a 5-minute Stick-figure Explainer for YouTube on why planes fly, in chapters.

Make a Caption News reel about this report: <link>. Sound off has to work.

Make a 3-minute 3D Map Explainer for YouTube on how the Panama Canal works.

GroupFolderWhat the picture looks like
Edit & documentarystyles/01-edit-documentary/real footage cut with big captions, newsreels, montage, broadcast graphics
Painting & illustrationstyles/02-painting/crayon, watercolour, oil, pen and wash, coloured pencil
Sketch & boardstyles/03-sketch-board/whiteboard, chalk, marker, sketchnotes, stick figures, tablet sketches
Traditional craftsstyles/04-traditional-crafts/ink wash, shadow puppets, ukiyo-e, paper-cut, stained glass, gold leaf
Printstyles/05-print/risograph, halftone, woodcut, engraving, screen print, pop art, collage
Graphic & typestyles/06-graphic-type/Swiss grids, spy titles, art deco, pictograms, Bauhaus, scale zooms, text flash cuts
Line diagramsstyles/07-line-diagrams/blueprints, isometric infographics, teardowns, cutaways, manuals, pen plotter
Cartoonsstyles/08-cartoons/1930s rubber hose, 80s cel anime, sitcom toons, comic history, manga panels
Gamesstyles/09-games/16-bit RPG, 2.5D pixel art, microgames, game shows
Film & tapestyles/10-film-tape/CRT terminal, silent film, found footage, scratch film, double exposure
Screens & interfacesstyles/11-screens/keynotes, live app screens, HUDs, desktop flythroughs, spreadsheets, 1-bit dither
Light & linesstyles/12-light-lines/circuit traces, fireflies, flow lines, an oscilloscope beam, neon tubes
3D & miniaturestyles/13-3d-miniature/brick toys, pop-up books, tilt-shift towns, low-poly islands, glass, dioramas
Instruments & evidencestyles/14-instruments/satellite, LiDAR, microscope, archive, X-ray, radar, thermal camera
Data & mapsstyles/15-data-maps/data stories, white models, physical data, transit maps, 3D maps, simulations

For YouTube explainers, start with the sketch-and-board group (whiteboard, stick figures, tablet sketch, sketchnote, marker on paper, chalkboard), Scale Zoom, 3D Map Explainer, Isometric Infographic, Line-Art Cutaway and Data Storytelling: they are built to hold a long explanation.

Workflow

Work in a run folder from edit-video (scripts/new-edit.mjs), one folder per video under work/<name>/.

  1. Brief. Style, topic, format (reel or YouTube), length, voice, and whether the user has their own footage, photos, data, logo or music. Ask once.
  2. Read the style, then write TREATMENT.md with directing: three structures, the one you chose, the opening frame by frame, a shot list (and chapters for YouTube), a cue map and a sound table.
  3. Lock the voice with edit-video's scripts/voice.py. Its word timings are the clock for every cut, caption and sound.
  4. Prove the look: two or three style frames drawn with the real code, checked at the format's real size against the STYLE.md. A storyboard only if the user asked.
  5. Build the composition with building: DOM and GSAP, or canvas and WebGL drawn from t, with a cues.json of one sound per visible event.
  6. Render with hyperframes render (a long YouTube video chapter by chapter), then mix and review with edit-video (sfx-sync.mjs, the opening, the must-read area, reading time, −14 LUFS).

Make a style yours

Change a library style instead of starting blank. Sort what you change into three:

  • Safe to change: palette, typefaces, pace, sound, which native moves you use.
  • Changes its character: materials and rendering, the colour logic, camera grammar.
  • Can't change: the style's essence (section 1). Change that and it's a different style.

Make at most five changes at a time, render one style frame next to the original, then a 15-second test. Save a kept style next to your videos as styles/<your-style>/STYLE.md, in the same shape, so later videos can use it by name.

Pitfalls already paid for

  • A title card or logo on the first frame loses the scroll. Every style's formats say what goes first instead.
  • A 16:9 composition shrunk into the middle of a 9:16 frame reads as a letterbox. Stack the layout; let pictures bleed.
  • In a reel, text right of x 900 or below y 1248 sits under the app's buttons and caption; on YouTube, the bottom 12 % sits under the player's controls.
  • A long YouTube video at one pace loses people halfway. Chapters and a re-hook every 30–60 s keep them.
  • A style's rules are a vocabulary, not a script: the opening, the ending and the signature shot come from the topic, not from the style's own examples.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.