by YeJe-cpu
An AI editor that watches its own cut: talking-head / AI-avatar video → auto-edited short video. 网感口播精剪：数字人/真人口播 → AI 自动精剪成网感动效短视频 → 可选剪映分层草稿
# Add to your Claude Code skills
git clone https://github.com/YeJe-cpu/SeeCutSee how SeeCut compares with popular alternatives.
SeeCut is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by YeJe-cpu. An AI editor that watches its own cut: talking-head / AI-avatar video → auto-edited short video. 网感口播精剪：数字人/真人口播 → AI 自动精剪成网感动效短视频 → 可选剪映分层草稿. It has 61 GitHub stars.
SeeCut's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/YeJe-cpu/SeeCut" and add it to your Claude Code skills directory (see the Installation section above).
SeeCut is primarily written in JavaScript. It is open-source under YeJe-cpu on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh SeeCut against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
An AI editor that watches its own cut — see, cut, loop.
数字人 / 真人口播 → AI 自动精剪成"网感"动效短视频 → 可选推成剪映分层草稿微调。 一个给 coding agent（Claude Code 等）用的 skill：它自己理解素材、自己去找真证据、自己编排和渲染，再用 AI 看视频给自己质检、自己迭代，最后交你肉眼终审。
| 数字人原片（HeyGen AV5 + 豆包配音） | SeeCut 自动精剪后 |
|---|---|
| https://github.com/user-attachments/assets/6c5c821a-bb8b-48f0-bead-5509425835ce | https://github.com/user-attachments/assets/4702dc92-8bb3-4199-acc0-31fd3d98ae96 |
两条都有声音，点开即可播放。右边的画面由 agent 全程自动完成（真证据截图、卡片、手绘圈注、人像在全屏和小窗之间调度），最后只在剪映里调了音量。原始文件在 Releases · demo-v1。
"精剪"不是去气口的粗剪；"口播"是有真人或数字人出镜；"网感"是画面有料、动效跟着每句话走。
zsh skill/seecut/scripts/preflight.sh，直到必需项全绿。skill/seecut 放进 ~/.claude/skills/（或你的 agent 的 skills 目录），新开对话：
/seecut 素材：<你的文件夹>
~/Downloads）+ 分层工程包 + 交付报告（每版改了什么、质检结论）。先样片、后全片。 长视频请先拿开头 20-30 秒跑一遍，确认风格和方向没问题再跑全片（素材超过 60 秒时 skill 也会先问你）。一次循环会出 2-3 版，每版都要渲染和 AI 看片对比，一条 25 秒样片目前约 1 小时，片子越长越久。
| 步骤 | 做什么 | 说明 |
|---|---|---|
| 第 1 步 · 数字人 | 口播文案 → 配音 → 数字人 → 竖屏口播 A-roll | 我们的做法：豆包配音 + HeyGen 数字人；工具可以换。有真人口播就跳过 |
| 第 2 步 · 网感口播精剪 | 素材理解 → 找真证据 → 编排 → HyperFrames 渲染 → AI 质检自迭代 → 交付 | 核心 |
| 第 3 步 · 剪映草稿 | 分层工程包 → 剪映里能打开、分轨微调的草稿 | 可选 |
目前只在 macOS 上实测过。克隆下来交给你的 agent，它会按这里和 preflight.sh 帮你装。
核心：精剪必需（preflight.sh 会逐项检查）
| 内容 | 说明 |
|---|---|
| 一个 coding agent | Claude Code 已实测；Codex、WorkBuddy 等能跑 skill 的 agent 理论上通用。要跑多轮自迭代，额度要充足 |
Antigravity CLI（agy） |
AI"看视频"的眼睛，没有它就没有自检和自迭代。用 Google 账号登录即可，免费账号有按周刷新的额度；Google AI Pro 会员额度更高、每 5 小时刷新（一条片子要调很多次，免费额度可能不够跑完，以官方说明为准）。坑：Google 账号底层地区要是支持地区（看账号设置，不看 IP）；国内要代理；无头调用要加 --dangerously-skip-permissions |
| HyperFrames CLI | 免费开源，npx -y hyperframes@latest 即用；代理要设成大写 HTTPS_PROXY |
ffmpeg、Node 22+（HyperFrames 要求）、Python 3、pip install faster-whisper、工程目录里 npm i playwright-core |
均免费，用于渲染、转写、截证据 |
数字人：要出数字人口播时需要（工具可以换，下面是我们实际用的）
| 内容 | 说明 |
|---|---|
| 配音：火山引擎·豆包语音 | 比数字人平台的内置音色自然。新应用有免费额度（我们开通时控制台显示约 2 万字符），以火山引擎控制台和官方最新文档为准 |
| 数字人：HeyGen 账号 + API 钱包 | AV4 照片驱动约 $0.04/秒，AV5 数字分身约 $0.12/秒 |
可选增强
| 内容 | 用途 |
|---|---|
| jianying-headless（第三方，非商业协议） | 生成剪映分层草稿 |
| 你自己的音效库 | scripts/sfx_ledger.py 一键建台账后自动配音效 |
| 一个生图工具 | 缺素材时自动生图补画面；任何生图模型都行，我们手边用的是 Grok CLI |
npx hyperframes 下载超时：代理没设成大写 HTTPS_PROXY。这套管线站在这些项目的肩膀上：
感谢这些项目的作者。本仓库包含的第三方组件与字体及其许可见 NOTICE。