VideoVerifiedInputs Converge To One Motion Clip
v1.0.9Use it to make a video.
npx skillloom-cli install sl-inputs-converge-to-one-motion-clip
Built for
Updated Oct 10, 2026
- Install guides
- Claude CodeCodex
- Generates with
- Host Agent LLM
- Uses tools
- PlaywrightFfmpeg
- In the package
7 files · 18.6 KB
- SKILL.md
- references/composition.html
- references/motion-contract.md
- references/scene-map.md
- references/style-guide.md
- scripts/render.mjs
- scripts/validate-video.mjs
Similar skills
VideoVerified
VideoVerifiedChapter Title Stamp Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLI
VideoVerifiedQuote Underline Portrait Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLI
VideoVerifiedEcho Burst Statement Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLI
VideoVerifiedRadial Branch Reveal Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLI
VideoVerifiedRatio Bars Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLI
VideoVerifiedConvergence To Evidence Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLI
VideoVerifiedQuote Sliders Motion Clip
Works with: Claude Code, Cursor, OpenCode, Codex, Gemini CLIWhat you get
- Trapezoid stages whose widths narrow step by step, each with the rule it applies
- A cohort bar that descends through the funnel and shrinks at every stage
- A drop chip at each stage that actually cuts something, carrying the figure it cut
- A locally rendered wide MP4 of 6 to 12 seconds, plus the editable source
Good fit
- Editorial transitions between a week of research and the one point worth making
- Educational posts that reduce many sources to a single claim
- Short publishing clips with no need for filmed footage
Not for
- A full animated explainer with multiple arguments
- A footage montage or product demonstration
Try this prompt
You say
/sl-inputs-converge-to-one-motion-clip Make an 11.6-second 16:9 clip for a fictional research desk about this week's inputs. Three stages, widest first: GATHERED 312 (“every clip, note and table the week pulled in”); SOURCED 74 (“cut anything without a date and a named publisher”, drops 238); ONE CLAIM 1 (“kept only what one sentence carries”, drops 73). Resolve to “312 inputs in. One claim out.” and close on “Many inputs. One claim.”

Then try
Use my own three stages and counts, keeping the same taper and timing.
Give the final claim a full extra second and shorten the opening title.
What you provide
- Three to five stages, widest first, each with its filtering rule
- The count that survives each stage and the count it drops
- The single claim the narrow end resolves into
Requirements
- Node.js 20+
- FFmpeg
- Chromium (npx playwright install chromium)
Running cost
No external services or generation models. Renders locally.
Verification
5.0/ 5
Render score
Verified
| Real render check | Verified | 5.0 / 5 | 67s |
|---|
FAQ
Can I change the stages or the counts after rendering?
Yes. Stage labels, rules, surviving counts, dropped counts and the final claim are plain values in the HTML/CSS source; re-render locally.
What if a stage drops nothing?
Leave its dropped figure blank. The engine shows a chip only where something is cut, so the first stage never gets one.
How many stages work best?
Three or four. Each stage needs about a second to read its rule and its count, and the clip runs 6 to 12 seconds.