by xianyu110
12 个电商短视频 Agent Skills:3秒钩子、卖点分镜、图生视频提示词、edge-tts 配音、字幕花字、ffmpeg 自动成片、抖音/TikTok Shop/视频号导出、直播切片、多语言配音、封面A/B · Claude Code / Codex / Cursor
# Add to your Claude Code skills
git clone https://github.com/xianyu110/ecommerce-video-skillsGuides for using ai agents skills like ecommerce-video-skills.
ecommerce-video-skills is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by xianyu110. 12 个电商短视频 Agent Skills:3秒钩子、卖点分镜、图生视频提示词、edge-tts 配音、字幕花字、ffmpeg 自动成片、抖音/TikTok Shop/视频号导出、直播切片、多语言配音、封面A/B · Claude Code / Codex / Cursor. It has 55 GitHub stars.
ecommerce-video-skills'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/xianyu110/ecommerce-video-skills" and add it to your Claude Code skills directory (see the Installation section above).
ecommerce-video-skills is primarily written in Python. It is open-source under xianyu110 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 ecommerce-video-skills 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.
One product photo → a ready-to-post 9:16 ecommerce video. 13 Agent Skills for hooks, viral remake, storyboards, edge-tts voice-over and a local ffmpeg render — no API key. 13 个电商短视频 Agent Skills · 一张商品图 → 带配音、字幕、BGM 的 9:16 带货视频 13 Agent Skills for ecommerce short videos — 抖音 · TikTok Shop · 视频号 · 小红书 · 淘宝主图视频 · Amazon
中文 · English · 效果 Demo · Skills · Related
👆 输入只有一张手机随手拍的商品图;配音、字幕花字、转场、BGM 全由 scripts/assemble.py 本地渲染(14.6 秒,1080×1920,无需 API Key)· 下载 MP4 👆 Input: one phone photo. Voice-over, captions, transitions and BGM rendered locally by scripts/assemble.py — no API key.
npx skills add xianyu110/ecommerce-video-skills
把「写脚本 → 找钩子 → 分镜 → 图生视频 → 配音 → 字幕花字 → 剪辑 → 导出 → 封面」拆成 13 个可单独安装的 Agent Skill。丢几张商品图给 Claude Code / Codex / Cursor,它会写钩子和分镜、出多模型镜头提示词,并且真的在本地把视频剪出来。
💡 怎么区分同名项目:本仓库的成片步骤是本地 Python + ffmpeg 渲染(edge-tts 配音免 Key),不需要任何 API Key、不依赖云端视频服务;图生视频模型只是可选项。 Not to be confused with other repos of the same name (e.g.
iart-ai/ecommerce-video-skills): this one renders the final video locally with ffmpeg — no API key, no cloud service required.
和一堆提示词的区别:
storyboard.json → edge-tts 逐句配音 → Ken Burns 镜头 → xfade 转场 → ASS 字幕/花字 → BGM 自动闪避 → 两遍响度标准化(≈ -14 LUFS)→ MP4 + SRT。纯 Python + ffmpeg,无需剪辑软件、无需 API Key。line_en 等字段,--lang en 直接渲染英文版(东南亚、日韩、拉美音色已列好)。npx skills add xianyu110/ecommerce-video-skills # 一行安装(skills CLI)
# 或在 Claude Code 里:/plugin marketplace add xianyu110/ecommerce-video-skills
手动安装:
git clone https://github.com/xianyu110/ecommerce-video-skills.git && cd ecommerce-video-skills
mkdir -p ~/.claude/skills && cp -r skills/* ~/.claude/skills/ # Claude Code(或项目内 .claude/skills/)
mkdir -p ~/.codex/skills && cp -r skills/* ~/.codex/skills/ # Codex
mkdir -p .cursor/skills && cp -r skills/* .cursor/skills/ # Cursor(项目内)
pip install -r requirements.txt # pillow numpy edge-tts;另需 ffmpeg(含 libass)
每个 skill 文件夹自带所需脚本,互不依赖,只复制需要的即可。
python scripts/assemble.py examples/aura-bottle/storyboard.json -o out/aura.mp4 # 中文版
python scripts/assemble.py examples/aura-bottle/storyboard.json --lang en --voice en-US-AndrewNeural -o out/aura-en.mp4 --build build/en
python scripts/export_platform.py out/aura.mp4 --platform douyin tiktok-shop channels xiaohongshu taobao-main amazon
python scripts/cover_ab.py out/aura.mp4 --images build/still00.png build/still03.png --titles "冰水撑不到中午?|早上的冰,晚上还在"
用 viral-remake 拆这条抖音爆款,换成我的保温杯复刻一条 15 秒分镜
用 hook-3s-script 给这个保温杯写 8 个前三秒开场,标出推荐 Top 3
用 selling-point-storyboard 按 ./photos 里的图做一条 15 秒抖音分镜,并输出 storyboard.json
用 ffmpeg-auto-assemble 把 storyboard.json 渲染成片,然后导出抖音和小红书版本
把这条视频本地化成美国英语和印尼语版本(multilingual-dubbing)
这是 3 小时直播录像和字幕,帮我切 8 条 30 秒高光(live-stream-clips)
商品图 ──► hook-3s-script ──► selling-point-storyboard ──► storyboard.json
│ ▲ │
│ └── viral-remake(爆款链接/转写 → 拆 5-beat → 复刻分镜)
└► image-to-video-shots ◄── multi-model-shot-prompts(可选:任意视频模型出镜头)
▼
ai-voiceover-edge-tts ─► subtitles-and-text-effects ─► ffmpeg-auto-assemble ─► MP4 + SRT
│
platform-spec-export ◄──────────────────────────────────────┤
cover-title-ab ◄────────────────────────────────────────────┤
multilingual-dubbing(--lang en / id / th …)◄──────────────┘
直播录像 ─► live-stream-clips ─► 9:16 切片 开箱/评测 ─► unboxing-comparison-review
多图示例:6 张商品图 → 17 秒成片(examples/aura-bottle):
| 中文成片(帧) | English version (frames) |
|---|---|
| ▶ MP4 中文 · 仓库内 docs/media/aura-bottle-9x16.mp4 | ▶ MP4 English |
封面 A/B(cover-title-ab,用干净底图 + 3 种标题/版式):
规格:1080×1920 · 30fps · H.264 + AAC · ≈ -14 LUFS · 17.0 秒 · 配音 edge-tts
zh-CN-YunxiNeural/en-US-AndrewNeural· BGM 由make_bgm.py实时合成(免版权)。 商品「AURA」是 ecommerce-image-skills 用 AI 生成的虚构商品,「24h 保冰」等卖点仅作演示。
| Skill | 中文说明 | 脚本 |
|---|---|---|
image-to-video-shots |
主图转视频:商品图 → 图生视频镜头提示词(运镜/动作/约束)+ 商品一致性锁 + 本地 Ken Burns 兜底 | assemble.py |
viral-remake |
爆款拆解复刻:链接/转写 → 8 维拆解 + 5-beat → 3 个复刻角度 → 可渲染 storyboard.json | — |
hook-3s-script |
3 秒钩子脚本:9 种钩子类型 + 2026 高完播角度,一次出 6–10 版(台词+画面+花字),广告法自检 | — |
selling-point-storyboard |
卖点分镜表:15/30/60 秒结构公式,一镜一信息,直接输出可渲染的 storyboard.json | — |
multi-model-shot-prompts |
多模型镜头提示词适配:通用骨架 + 运镜词表 + 文生/首帧/首尾帧写法,不编造价格/参数 | — |
ai-voiceover-edge-tts |
AI 口播配音:edge-tts 逐句合成、去静音、时长统计、缓存;口播稿改写规则与音色推荐 | tts_edge.py |
subtitles-and-text-effects |
字幕与花字:ASS 四种样式(字幕/弹出花字/贴纸/CTA)、关键词高亮、安全区 | make_subs.py |
ffmpeg-auto-assemble |
ffmpeg 自动拼接成片:一条命令出片(配音、转场、字幕、BGM 闪避、响度标准化) | assemble.py 等 5 个 |
platform-spec-export |
TikTok Shop / 抖音 / 视频号规格导出:9:16 / 3:4 / 1:1 / 16:9,模糊填充或裁切,安全区预览 | export_platform.py |
live-stream-clips |
直播切片:时间戳 / 字幕关键词 / 停顿提议三种选段,批量导出竖屏切片 | live_clip.py |
unboxing-comparison-review |
开箱 / 对比评测模板:结构、拍摄清单、公平对比规则、storyboard 示例 | make_cards.py |
multilingual-dubbing |
多语言本地化配音:本地化改写 + 各市场音色 + --lang 一键渲染多语言版本 |
assemble.py |
cover-title-ab |
封面与标题 A/B:自动挑清晰帧/干净底图 + 多标题版式,对比图与测试记录表 | cover_ab.py |
不需要。这些 skill 在任何能读 SKILL.md 的 agent 里都能用,本地剪辑脚本不需要任何 API Key;只有在你想用图生视频模型生成镜头时才需要对应模型的账号。第三方模型/接口入口见文末 Related。
详见 CONTRIBUTING.md。欢迎 PR:新的平台规格(附官方链接)、更好的钩子/分镜模板、更多脚本。请保持每个 skill 自包含(skills/<name>/SKILL.md + scripts/);修改 scripts/ 后运行 python tools/sync_scripts.py 同步到各 skill。
MIT © xianyu110。内容为原创编写;参考项目(MoneyPrinterTurbo · MIT、shuohao-skills · Apache-2.0、reelbench-skills · Apache-2.0、ffmpeg-skill · MIT 等)只借鉴了结构思路,未复制文本或代码,详见 docs/credits.md。edge-tts 为 LGPL-3.0 依赖;FFmpeg 由用户自行安装。平台规则经常变化,上传前请以各平台官方说明为准。
Thirteen installable Agent Skills that take an ecommerce product from a few photos to a finished 9:16 short video — hooks, storyboards, model-agnostic image-to-video prompts, free edge-tts voice-over, styled subtitles, a working local ffmpeg render pipeline, platform exports (TikTok Shop, Douyin, WeChat Channels, Xiaohongshu, Taobao, Amazon), live-stream clipping, unboxing/comparison templates, multilingual dubbing and cover/ti