SkillFoxx

Ecommerce Video Skills

13 skills for short product videos: hooks, storyboards, edge-tts voice-over and a local ffmpeg render in 9:16 with no API keys

  • 139 ★ on the repository
  • Last commit
  • MIT
Skill

Extended access

Why this level

  • Runs local scripts and ffmpeg and writes clip files and intermediate frames into the working folder
  • edge-tts voice-over uses the network and the clip breakdown skill accepts links to other people's videos
All reasons and checks

xianyu110/ecommerce-video-skillsAuthor:xianyu110

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add skills/ecommerce-video-skills

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Runs local scripts and ffmpeg and writes clip files and intermediate frames into the working folder
  • edge-tts voice-over uses the network and the clip breakdown skill accepts links to other people's videos

This is third-party code. Review the repository files before installing.

What it does

Ecommerce Video Skills splits the path from script to finished clip into 13 separately installable skills. From one or several product photos you get a vertical 9:16 video with voice-over, captions, music and transitions. The skills include hook-3s-script for the first three seconds, selling-point-storyboard for a benefit-based storyboard, image-to-video-shots and multi-model-shot-prompts for model prompts, ai-voiceover-edge-tts for voice-over, subtitles-and-text-effects, ffmpeg-auto-assemble, platform-spec-export, cover-title-ab, multilingual-dubbing, live-stream-clips, unboxing-comparison-review and viral-remake. Assembly runs as a local Python and ffmpeg pipeline with no API keys: storyboard.json becomes per-line voice-over through edge-tts, Ken Burns camera moves, xfade transitions, ASS captions and titles, music with auto-ducking and two-pass loudness normalization around −14 LUFS. The output is MP4 and SRT. scripts/export_platform.py fits the clip to Douyin, TikTok Shop, WeChat Channels, Xiaohongshu, Taobao and Amazon, and scripts/cover_ab.py prepares cover variants for A/B testing. The viral-remake skill takes a Douyin or TikTok link or a transcript, breaks the clip into five beats, gives three angles for your product and builds a render-ready storyboard. Image-to-video generation is optional and runs on your own model.

Who it is for. Marketplace sellers and studios that make short product videos for Douyin, TikTok Shop, Xiaohongshu and Amazon from a few photos.

Good fit when

  • You have only product photos and need a 9:16 clip with voice-over, captions and music in one pass
  • The first three seconds of a clip are weak and you need several hooks
  • You need a hit clip broken down into five beats and remade for your product
  • One clip must ship to several platforms with different sizes and limits
  • The clip is needed in another language: multilingual voice-over through edge-tts

Not a fit when

  • You need realistic video generated by a model: the pipeline builds a clip from your photos and frames
  • You need a finished edit of live footage: the skill is built on still frames with the Ken Burns effect
  • You need an exact copy of someone else's clip: viral-remake breaks down structure and finds a new angle
  • The machine has no ffmpeg with libass: captions will not render without it

Example request

Take three thermos photos from the photos folder and build a 15 second 9:16 clip: hook, benefit storyboard, edge-tts voice-over and an export for TikTok Shop

Limitations

  • Python, ffmpeg with libass and the pillow, numpy and edge-tts packages from requirements.txt are required; edge-tts voice-over needs no key.
  • Image-to-video generation is optional and runs through your own model; the set has none built in.
  • The hit-breakdown skill takes a link or transcript and the remake is built for your product: the author does not intend copying someone else's video as is.
  • The repository shares its name with other projects and the author asks not to confuse it, for example with iart-ai/ecommerce-video-skills.

How to disable. Delete the skill folders from the agent skills directory, for example ~/.claude/skills, and the repository clone. Finished clips stay in your output folder.

FAQ

How do I install Ecommerce Video Skills?

Run npx skills add xianyu110/ecommerce-video-skills. For local assembly add pip install -r requirements.txt and install ffmpeg with libass.

Do I need API keys to build a clip?

No. edge-tts voice-over, transitions, captions and music are assembled locally in Python and ffmpeg. A key is needed only if you connect a model for image-to-video yourself.

How do I assemble a clip from a storyboard?

Run python scripts/assemble.py examples/aura-bottle/storyboard.json -o out/aura.mp4. For an English version add --lang en --voice en-US-AndrewNeural.

Which platforms does the export support?

python scripts/export_platform.py out/aura.mp4 --platform douyin tiktok-shop channels xiaohongshu taobao-main amazon fits the clip to these platforms.

What does viral-remake do?

It takes a Douyin or TikTok link or transcript, breaks the clip into five beats and eight dimensions, offers three angles and builds a render-ready storyboard for your product.

Discussion

No questions or reviews yet.

Skills

12 skills that turn one product photo into an Amazon main image, lifestyle scene, A+ module, sale banner and Xiaohongshu cover

PluginExtended access176

Free Claude Code skills for short-form video: scripts, local editing, translation with a cloned voice and publishing.

SkillExtended access156

A skill for e-commerce visuals: builds prompts and generates product hero shots, cards and ad images from tested recipes

SkillExtended access1.7K

A skill that turns a topic into a narrated 4K video: research, script, local TTS, Remotion composition and an FFmpeg mix

SkillExtended access1.7K
Needs a VPN

A skill that turns an idea, a link or a screen recording into finished video: HeyGen avatar, b-roll, captions and quality checks

SkillBroad access343

README in short

The README is bilingual: first the short promise that one photo becomes a 9:16 clip, then install, assembly, export and cover scripts, a skills diagram with a chain from hook to export, a table with the description and triggers of each skill, a pointer to the sister repository for images and contribution advice.

README badge

Original SKILL.md

---
name: hook-3s-script
description: "3 秒钩子脚本：为电商短视频写前 3 秒开场（台词 + 画面 + 花字），按痛点/反差/结果先行/提问/数字/场景代入等钩子类型一次给出多版，并做合规自检。Use when the user asks for a hook, 开头, 黄金 3 秒, 前三秒, opening line, scroll-stopper, 完播率, or wants to rewrite a weak opening for 抖音 / TikTok Shop / 视频号 / Reels."
---

# 3 秒钩子脚本 · 3-Second Hook

前 3 秒决定用户会不会划走。这个 skill 只做一件事：**给同一个商品写 6–10 个不同类型的开场，每个都带台词、画面、花字**，方便直接 A/B。

> Part of [ecommerce-video-skills](https://github.com/xianyu110/ecommerce-video-skills)

## 输入 Inputs

- 商品是什么、给谁用（人群 + 场景）
- 1–3 个真实卖点（必须是用户确认过的事实）
- 平台与语言（中文 / 英文 / 双语）
- 可选：竞品常见说法、差评里的痛点、已有素材（图 / 视频）

## 步骤 Steps

1. **找一个最痛的场景**：从「用户在什么时候会烦」出发，不从「产品有多好」出发。
2. **按下表钩子类型各写 1 条**，每条 ≤ 18 个汉字（中文语速约 4–5 字/秒，3 秒说完）。
3. **配画面**：第一帧必须出现商品或痛点画面，别用 logo 片头。
4. **配花字**：花字 ≤ 10 字，是台词的「标题版」，不是逐字重复。
5. **合规自检**（见清单），删掉无法证明的说法。
6. **输出**：表格 + 推荐 Top 3 + 一句话说明为什么。

## 钩子类型 Hook patterns

| 类型 | 公式 | 示例（保温杯，演示） |
|---|---|---|
| 痛点提问 | 你是不是也…？ | 冰水不到中午就变温了？ |
| 反差 | 看着像 A，其实 B | 看着像普通水杯，晚上还能倒出冰块 |
| 结果先行 | 先给结果画面，再解释 | （先拍晚上倒出冰块）这是早上装的冰 |
| 数字 | N 个 / N 秒 / N 小时 | 通勤包里放了 8 小时，一滴没漏 |
| 场景代入 | 当你在 X 的时候 | 健身完想喝口冰水的时候 |
| 错误示范 | 别再这样做了 | 别再往包里塞会漏的杯子了 |
| 对比 | 左边 vs 右边 | 左边普通杯，右边这只，同时装冰 |
| 第一人称安利 | 我… | 我换了三个杯子，最后留下这只 |
| 悬念 | 最后一个最关键 | 选保温杯看这 3 点，第 3 点最多人忽略 |

Report a problem