speechlab0210/tutorial-video-skill
General tutorial-video skill: coherent, accurate, editable。
见原仓库 README
OPUS 5.5 LAB
A curated gallery of Claude Opus 5.5 video and 3D cases with traceable original posts, published prompts, and tool notes.
Hover a card for a muted preview; playback connects to the source video service.
General tutorial-video skill: coherent, accurate, editable。
见原仓库 README
Product URL → 16:9 launch film (EN/KR): real UI capture, one-take voice-over, beat-locked motion。
真实 UI 录制 + 代码动效
Give the agent a website, some footage, or a topic — get back a finished short video。
见原仓库 README
3 master skills: edit video, make bangers, build anything。
见原仓库 README
Editing judgment for AI agents: pacing, cuts, captions, audio, color, thumbnails。
见原仓库 README
Make and edit videos with Claude Opus 5.5: product films, ads, talking heads。
见原仓库 README
Record + auto-edit web-app short videos (walkthrough / QA)。
录屏 + 自动剪辑
Whiteboard talk editing: rough cut, eased zooms, color。
见原仓库 README
30 narrow, focused editing skills + presets (Apple-Standard look)。
见原仓库 README
9:16 Chinese crypto/AI explainers for Twitter KOLs, with personal branding。
见原仓库 README
Chinese short videos → English dubbing + English subtitles (MiMo TTS/ASR + Remotion)。
MiMo TTS/ASR + Remotion
14 Codex skills: video comparison, hook selection, mashups。
见原仓库 README
Agent skill + MCP: search 24fps for AI motion graphics / prompts / editing presets。
24fps MCP 检索
AI-assisted motion-design tools + skills + prompts (Three.js/GSAP/HyperFrames/Remotion)。
Three.js/GSAP/HyperFrames/Remotion
Teaches your AI to watch videos — learn, absorb, imitate, or give visual feedback。
多模态视觉 + ffmpeg 抽帧
Turns videos and courses into evidence-grounded Agent Skills。
ffmpeg 抽帧 / 转写
Lets Claude watch & understand local videos via ffmpeg + local Whisper; offline, drops into `~/.claude/skills/`。
ffmpeg + 本地 Whisper
Any video link → one typed JSON: summary / chapters / people / brands / emotion arc / transcript。
视频理解 API + skill
Reverse-engineers any video (YouTube/TikTok/Reels/X) into a reusable production blueprint, with evals。
视频理解 + blueprint 输出
Measure videos on macOS: UI animation timing and easing back into CSS, scene splitting。
帧分析 + CSS 输出
Bilibili/YouTube video → illustrated tech blog: frame extraction + dedup, AI subtitles, verifiable transcript, timestamp links。
抽帧 + 转写
Douyin/Bilibili/Xiaohongshu/YouTube → transcript + frames (evidence-linked)。
转录 + 抽帧
Gives agents video input: watch / index / ask / iterate (MCP/CLI/REST/Python)。
视频输入通道(MCP/CLI/REST)
The lab indexes cases and links from these public repositories.