Skip to main content

edit-video

Edit a raw talking-head video through transcription, silence trimming, mistake review, visual storytelling, motion graphics, and verified HyperFrames rendering. Use for a complete edit; route a single requested operation directly to its specialist skill.

ソース情報

リポジトリ
nateherkai/hyperframes-student-kit
ソースの最終更新活動
2026年9月8日 18:40
検出された SKILL.md の言語
英語
スター
1,312
フォーク
386

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

SKILL.md を表示中

SKILL.md
ソースの指示 · 読み取り専用プレビュー
name
edit-video
description
Edit a raw talking-head video through transcription, silence trimming, mistake review, visual storytelling, motion graphics, and verified HyperFrames rendering. Use for a complete edit; route a single requested operation directly to its specialist skill.
# Edit a video Keep each project in `video-projects/<slug>/`, preserving the source recording. Read the workspace guide and the project's brief or DESIGN.md. If a direction is missing, propose a style from `style-library/registry.json` based on the footage. 1. Inspect duration, streams, and source resolution with ffprobe. Copy the source into the project's assets or reference an explicitly provided local path. 2. Read `docs/TOOLS-AND-API-KEYS.md` before service setup. Reuse a matching word transcript. Otherwise use the student's chosen provider; Nate defaults to `node scripts/transcribe-elevenlabs.mjs <source>`. OpenAI Whisper and local Whisper are supported workflow alternatives after transcript normalization. Check available credentials or local dependencies. Explain uploads and costs before any service call whose authorization is still missing. 3. Read `../cut-silences/SKILL.md`. Produce an EDL and a retimed transcript, and render the silence pass when editing is authorized. 4. Read `../cut-mistakes/SKILL.md`. Inspect candidates in context and preserve intentional emphasis. Record the reviewed cuts. If no mistakes need cutting, carry forward the silence output. Match every transcript to its actual video. 5. Read `../video-storytelling/SKILL.md` for the visual arc and `../hyperframes-video-beats/SKILL.md` for overlays. Write a beat sheet with transcript anchors, one visual idea per beat, and deliberate callbacks. 6. Read `MOTION_PHILOSOPHY.md`, `../hyperframes/SKILL.md`, and relevant GSAP references before authoring. Use `../style-library/SKILL.md` to select or adapt a card. Localize every asset used by the composition. 7. Run preflight and HyperFrames lint from the project, plus the transcript-sync validator for anchored sub-compositions. Review Studio before the draft. 8. Render a draft, inspect encoded frames and transitions, and listen across cut boundaries. Check face framing, text legibility, black frames, and A/V sync. Resolve failures before the final render. Follow any review approvals already provided; do not repeat approval requests for the same authorized action. Deliver the final MP4 path, edit decisions, retimed transcript, composition, and `VERIFY.md` describing what was checked and any remaining limitations. An automated detector proposes edits; editorial judgment remains necessary.
GitHubで見る