All skills
hoodini avatar

/yuv-video-director

@f412198

Yuval's all-in-one AI video pipeline. Turns an idea/script into a finished, on-brand MP4 by orchestrating HyperFrames (HTML→deterministic video render), Lottie (branded motion graphics), ManimCE (math / neural-network / concept animations), and a transcribe→approve caption flow — all wrapped in the YUV.AI Neon Phoenix brand via a frame.md. Use whenever Yuval wants to make, edit, or explain something as a video: promo, explainer, launch, social reel, "make a video about X", "explain X as a video", "neural network animation", "turn this into a video", captioned tutorial, 16:9 or 9:16. Triggers: video, explainer, promo, reel, manim, lottie, hyperframes, animation, "make a video", "explain ... as a video", מצגת וידאו, סרטון, הסבר וידאו. Routes each beat to the right engine, wraps in brand, self-verifies, and renders.

Use this Skill: https://skilld.dev/gh/hoodini/ai-agents-skills/yuv-video-director

This session only. Nothing lands on disk.

referencesediting.md

≈846 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Editing — cut it like a teaser, not a slideshow

The difference between "AI made a video" and "a producer made a teaser" is the edit: shot length, rhythm, hard cuts, kinetic slams, a montage, a build → reveal. Default to this for any social/promo/teaser; relax to longer holds for explainers.

Shot rhythm

  • Hook in the first ~1s. Open on the boldest thing (logo slam + particle burst), not a slow fade.
  • Short shots. Montage shots 0.4–0.9s; statement slams 0.6–0.9s; a "rest" beat (terminal, code, the Manim clip) can run 3–5s so the eye recovers. Alternate fast/slow — that contrast IS the pacing.
  • Hard cuts are the default. A teaser is mostly hard cuts (clip ends → next clip starts). Don't put a transition on every cut — that reads slow. Reserve effects for hits.
  • Stabs on the hits. A quick glitch + chromatic flash (~0.08–0.16s) on the big beats (logo, each engine slam, the brain beats, the title) punctuates without slowing.

Kinetic slams (the teaser staple)

Single word, full-screen, Anton uppercase, scale-punch in, hard cut out (no exit anim):

const slam = (sel, t) => tl.from(sel, { scale:0.35, opacity:0, duration:0.2, ease:"back.out(2.2)", overwrite:"auto" }, t);
slam("#e1 .slam", 6.7);   // "HYPERFRAMES"

Each slam is its own ~0.6–0.9s clip. Rainbow-grad the brand words, neon-glow the punch words. Alternate color/size between consecutive slams so they don't blur together.

Structure that works (the shipped teaser)

0.0  HOOK montage — YUV.AI slam+burst · code flash · palette flash · "VIDEO" slam   (4 shots, ~0.5s each)
2.25 PROMISE — "ONE PROMPT" slam → terminal types the command + agent steps          (rest beat ~3.7s)
6.7  ENGINES — HYPERFRAMES · MANIM · LOTTIE · FRAME.MD                                (4 slams, ~0.85s)
10.2 BRAIN  — "TRAIN" / "FREEZE" slams → Manim clip beat                              (2 slams + 5s clip)
16.4 PROOF  — code editor reveals the real HTML                                       (rest beat ~3.6s)
20.0 CLIMAX — "LET'S FLY HIGH" rainbow slam + burst + CTA → fade                      (~7s payoff)

Build → rest → build → rest → payoff. Stabs at: every slam-in, the clip-ins, the title.

Two-track caption hint

Don't put exit anims on shots (rule). The hard cut between clips IS the exit. Only the final clip fades.

Real WebGL shader transitions (optional, heavier)

The catalog's glitch, swirl-vortex, gravitational-lens, chromatic-radial-split, light-leak, cinematic-zoom etc. (npx hyperframes add <name>) install as showcase blocks (full demo compositions with their own content) — not drop-in A→B transitions. True frame-to-frame shader transitions use the @hyperframes/shader-transitions package API (capture from/to textures). For most teasers, the hand-built glitch + chromatic flash stab above is faster, deterministic, and reads just as punchy. Reach for the shader package only when a specific hero transition demands it.

Pacing checklist

  • Hook in ≤1s. [ ] No shot the eye can't read (slams ≥0.5s on screen). [ ] Fast/slow alternation.
  • Hard cuts default; stabs only on hits. [ ] One idea per shot. [ ] Climax is the longest hold.
  • Spot-check at 30fps that no slam samples mid-entrance as the only visible frame.

Source: SKILL.md on GitHub

1 warning3mo3 checks · Risk SAFE
  • Gen Agent Trust Hub3mo

    The yuv-video-director skill is a comprehensive and well-architected video production pipeline for AI agents. It orchestrates HyperFrames, Lottie, and ManimCE to create professional-grade video content with synchronized captions and branded motion graphics. The skill follows standard practices for asset management, using helper scripts to fetch legitimate resources like fonts and libraries from trusted sources. No malicious patterns, obfuscation, or data exfiltration risks were identified during the analysis.

  • Socket3mo

    No alerts

  • Snyk3mo

    Risk: MEDIUM · 1 issue

Signed by skilld at f412198. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Steadyupdated 4 months ago

README badge

README badge for hoodini/ai-agents-skills/yuv-video-director