Name: File-Image Toolkit
Author: Victor42

能力

你说一句，它干活

「把这些 png 转成 webp」→ 整个文件夹批量转，同名文件自动加序号
「把这张图缩到 800px」→ 等比缩放，不裁掉内容
「这几张图合成 GIF」→ 多帧合成，也能把 GIF 拆回帧序列
「从视频里提取音频」→ mp4 直接出 mp3
「把嵌套文件夹里的文件都提取出来」→ 压平到一层，文件名不丢
不用自己写命令：装好后用自然语言说，Agent 去调脚本

图片

格式互转：jpg / jpeg / png / webp / avif / gif / bmp / tiff 任意互转
按尺寸改：指定宽高，裁剪或填充
按边长缩放：设定最大边长，保持比例
裁头像：自动检测人脸，输出圆形或方形
去重：感知哈希找相似图，副本移入单独目录，保留最清晰那张
GIF：多帧合成，或拆成帧序列

音视频

音频互转：mp3 / wav / flac / aac / ogg / m4a / opus
视频互转：mp4 / avi / mov / mkv / webm 等，可转 GIF
提取音轨：从视频里取出音频
图片合成视频：多张图拼接，支持交叉淡入淡出

下载

公众号文章内嵌视频：按视频标题命名，可选最高清或全部清晰度

文件管理

嵌套文件夹压平到一层
按 CSV 批量重命名
单文件与整文件夹都支持

动手前会发生什么

首次运行自动检查 Python 3.8+ 与依赖包，缺了帮你装
只有音视频的活才需要 FFmpeg（4.3+），纯图片处理不需要
批量转换、压平、重命名可以先看一份计划（dry-run）再执行
同名文件自动加序号，不覆盖原文件

Capabilities

Say it, it runs it

“Turn these pngs into webp” → whole-folder batch, same-name files get numbered
“Scale this image to 800px” → proportional, nothing cropped
“Turn these images into a GIF” → multi-frame compose, or split a GIF back into frames
“Pull the audio out of this video” → mp4 in, mp3 out
“Flatten this nested folder” → one level, no filenames lost
No commands to write: install it, speak plainly, the agent calls the scripts

Images

Convert between jpg / jpeg / png / webp / avif / gif / bmp / tiff
Resize by exact width and height, crop or pad
Scale by longest edge, keeping the aspect ratio
Avatar crop: face detection, round or square output
Dedupe: perceptual hash finds near-duplicates, moves copies to a separate folder and keeps the sharpest one
GIF: compose frames into a GIF, or split one back into frames

Audio and video

Convert mp3 / wav / flac / aac / ogg / m4a / opus
Convert mp4 / avi / mov / mkv / webm and more; video to GIF
Extract the audio track from a video
Build a video from stills, with cross-fades

Downloads

Videos embedded in WeChat articles: named after the video title, highest quality only or every quality

File management

Flatten a nested folder into one level
Batch rename from a CSV
Single files or whole folders

What happens before it touches your files

First run checks Python 3.8+ and the dependency packages, installing what's missing
FFmpeg (4.3+) is only needed for audio and video work; image-only tasks don't need it
Batch convert, flatten and rename can run a dry-run plan first
Same-name files get numbered; originals are never overwritten

给 AI 的安装指令

发给 AI。

请帮我安装 Victor42 的 文件图像工具包 skill：
1. 在终端运行：npx skills add greenzorro/victor42-skills --skill file-image-toolkit -g
2. 若没有 Node，可改为下载 https://cloud.victor42.work/skills/file-image-toolkit.zip，解压到当前环境的 skills 目录（文件夹名 file-image-toolkit），确认其中存在 SKILL.md
3. 完成后告诉我

Please install Victor42's File-Image Toolkit skill:
1. Run in your terminal: npx skills add greenzorro/victor42-skills --skill file-image-toolkit -g
2. If you don't have Node, download https://cloud.victor42.work/skills/file-image-toolkit.zip, extract into your agent's skills directory as file-image-toolkit, and confirm SKILL.md exists
3. Tell me when it's done

文件图像工具包

File-Image Toolkit

触发词

Triggers

能力

你说一句，它干活

图片

音视频

下载

文件管理

动手前会发生什么

Capabilities

Say it, it runs it

Images

Audio and video

Downloads

File management

What happens before it touches your files

给 AI 的安装指令