Communitygithub.com

LiamVisionary/hivemindos

Use when a user wants a polished product launch video with generated footage, motion-graphics overlays, reserved text zones, storyboard beats, text-safe composition, and final render QA.

hivemindos 是什么?

hivemindos is a Claude Code agent skill that use when a user wants a polished product launch video with generated footage, motion-graphics overlays, reserved text zones, storyboard beats, text-safe composition, and final render QA.

兼容平台~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/LiamVisionary/hivemindos/tree/HEAD/packaged-skills/optional/media/hivemindos/launch-video-hyperframes

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Launch Video Hyperframes

Create launch-video prompts and assembly instructions around text-safe motion frames.

Core Principle

Every shot must reserve a readable text zone before generation or filming. Do not add overlays after the fact onto busy footage and hope they fit.

Inputs

  • Product, audience, launch moment, CTA, proof points, and forbidden claims.
  • Duration, aspect ratio, platform, voiceover preference, and output format.
  • Existing footage, product screenshots, brand assets, or generated-media preference.

Workflow

  1. Write the launch spine: hook, product reveal, mechanism, proof, momentum, CTA.
  2. Create a beat board with shot duration, visual, text zone, overlay line, audio line, and transition.
  3. For each shot, reserve a clear third of the frame or another explicit safe zone.
  4. Generate footage or use supplied media. Verify the safe zone remains clean.
  5. Build overlays with consistent type, spacing, and motion. Keep text short enough for mobile.
  6. Assemble, then verify playback, readable overlays, audio sync, contrast, and platform dimensions.
  7. Deliver source prompts, render path, and known limitations.

Capability Map

  • Video generation: use the discovered best video capability for the requested style.
  • Motion graphics: use local code/HTML/Remotion/FFmpeg/GSAP when available rather than flattening everything into prompts.
  • QA: use video-render checks and screenshot sampling before claiming final.

Safety

  • Do not invent traction, revenue, customer logos, compliance claims, or endorsements.
  • Ask before uploading private screenshots or unreleased product footage.
  • Do not publish without approval and receipt.

Provenance

This is a HivemindOS-authored clean-room skill inspired by a public skill catalog. No unlicensed upstream text, JSON, or scripts are copied.

相关技能