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

animated-explainer

Skill75 starsChanged 39 days agoApache-2.0View on GitHub

Style for edit-video: a fully animated reel with no camera. A locked voice (yours, a free local voice or ElevenLabs), one lit colour field and one big hero card per spoken beat, the spoken line word-lit at the top, and an optional cartoon character as a cursor that drives the UI. Use for tool, product and idea explainers, two-voice conversations and story reels when the creator doesn't want to film.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Spec: Write build(T) in spec.py: one scene per beat starting on a line or a word (T.L(i), T.find('word', line=i)), its hero markup, the …

    line 54
  2. Distinctive:

    Look at every frame: text readable and inside the safe area, nothing overlapping, the character never on text.

    line 65
  3. Very specific:

    All from edit-video's SFX pack.

    line 91
  4. Distinctive:

    No bass hits, and leave one or two quiet beats.

    line 107
  5. Distinctive:

    Don't rebuild between the reviewed render and the delivery render.

    line 119

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Animated explainer

A style on top of edit-video, which owns the sound mix, safe areas and review. The whole reel is animation keyed to a voice: every scene starts on a spoken line, every card pops on the word that names it, and every sound lands on the frame that shows its event. No camera, no footage. Add the cartoon-character skill to put a lip-synced character in it.

What makes it look premium

  • The voice is the clock. Lock the voice first and key everything to its word times. Never time animation by ear or by reading speed.
  • A lit world, one field per beat. Each scene is a full-bleed colour field with a soft radial light (night, ultra, forest, clay, ink, and cream, lilac or peach for light beats). Neighbours differ, so the reel moves even with the sound off. Every scene pushes in slowly; nothing is ever a still.
  • One big hero object per beat, with real content: a calendar, a terminal, a chat, a receipt, a timeline. Size it from its content, about 864 px wide, never a small card in an empty field. Show rows still to come dimmed rather than leaving a card half empty. Label anything illustrative that looks like data.
  • Type: one heavy sans, one serif italic for the single accent word, one mono for labels. The hook and the spoken line are 84–112 px, at most three lines.
  • Motion: entrances rise or pop and are fully visible within 0.07 s, so they never arrive after their sound. Stamps land on their word. Counters roll once to their true value. Transitions come from the medium: an iris from where the cursor clicked, a scroll for a turn, a cut on the same field. No dissolves, no blank frame.
  • A character is small and busy. As a cursor badge it moves to what is being said, clicks it and rests beside it. It never covers text being read.
  • Frame 0 already shows the hook and the hero. Something changes every few seconds.

These are defaults. A creator's brand, palette and reference win; keep the method.

Workflow

Work in a run folder from edit-video (scripts/new-edit.mjs). Each reel is a folder work/<reel>/ that holds its spec.py, its voice/ and the built HyperFrames project. Start by copying the worked example.

  1. Script. One idea, 15–40 s. A line that starts with [Agent] (or the spec's GUEST name) is the second voice. Check every name, number and feature against its source; a first-person claim needs a real run.
  2. Voice. edit-video's scripts/voice.py work/<reel> --engine files|hyperframes writes voice/vo.wav and voice/timing.json (every word's start and end). Which engine, and how to clone your own voice for free, is in edit-video's voice options (references/voice.md). Listen to the takes: timings can't catch a mispronounced word. After this the words are locked; change only | phrase breaks and *accents*.
  3. Spec. Write build(T) in spec.py: one scene per beat starting on a line or a word (T.L(i), T.find('word', line=i)), its hero markup, the timeline JS using the helpers in assets/explainer.js (pop, rise, fade, nudge, count, type, draw, push), the cursor path, the character's poses, and one sound cue per visible event. The full contract is in the docstring of scripts/build.py.
  4. Build: python3 <skill>/scripts/build.py work/<reel>. It writes index.html, cues.json and assets/, and stops if the script and the voice disagree on a line's words.
  5. Look before rendering. hyperframes snapshot work/<reel> --describe false --at … at frame 0, each scene's first spoken word, every click and the last second. Look at every frame: text readable and inside the safe area, nothing overlapping, the character never on text. Fix and repeat.
  6. Render with hyperframes render work/<reel> --fps 30 --quality delivery --output previews/<reel>-v01.mp4.
  7. Sound. Mix the voice and the cues with edit-video's scripts/sfx-sync.mjs mix work/<reel>/cues.json --voice work/<reel>/voice/vo.wav --out work/mix-v01.wav --video previews/<reel>-v01.mp4 --mux exports/<reel>-v01.mp4, then check it with sfx-sync.mjs check … --strips reports/sync-v01 and look at the strips.
  8. Review and deliver with edit-video: frame 0 and the first 3 seconds, a 4 fps sheet of the whole reel, phone-size safe areas, loudness. Say what you could not judge, such as a listening pass.

Layout at 1080×1920

ZoneWhere
Spoken linex 128–928, from y 330. The hook can be bigger (BIG)
Hero cardx 108–972, y 650–1250; its text above y 1248
Cursor badge at restabout x ≤ 740, y ≥ 1290; arrive, click and leave within about 0.6 s when it clicks inside a card
BottomCan stay field: the platform's caption sits there

STYLE = 'rail' switches to white grid paper with the character's head riding a progress rail down the left edge and a chip naming each chapter; cards then start at x 340.

Sound map

All from edit-video's SFX pack. Gains in the spec are relative to each sound's suggested level.

Visible eventSound
Frame 1 of the hookui-pop
Card or panel arrives; button appearsui-pop
The cursor clicksui-click
A row, step or marker appearstick
Something lands or a stamp slamsthock
Approved, connected, donelatch
Progress or a counter runningratchet
Typingtyping-tick
Scene changezoom-swish, about 0.12 s before the line
Earned completion, onceclean-ding

No bass hits, and leave one or two quiet beats.

Pitfalls already paid for

  • A sound cued at a scene's very first frame plays before the transition shows its element; cue it once the element can be seen.
  • A cursor that leaves for its next click too early flies across the previous scene's card. Start a move after the new scene has opened.
  • A number word in the script ("six") and a digit in the transcript ("6") are matched; other mismatches are placed between their neighbours and reported. Fix the take or the script when that happens.
  • Element ids in a spec must be unique; prefix them with the scene number.
  • Don't rebuild between the reviewed render and the delivery render.

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.