← Back to Skills Marketplace
139
Downloads
1
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install deduplicate-images
Description
递归扫描全目录图片，自定义相似度，安全软删除至回收站，保留高清原图
README (SKILL.md)
🖼️ 图片智能去重专业版 | Image Deduplicator Pro
核心功能 | Core Features
✅ 递归深度扫描：根目录+所有子文件夹全量图片检索 ✅ 自定义相似度：支持 0~100% 任意阈值设置 ✅ 安全软删除：仅移动文件，不删除任何源文件 ✅ 智能择优保留：优先留存高分辨率、大体积优质图片 ✅ 全格式兼容：支持 JPG/PNG/WebP/BMP/TIFF ✅ 跨平台运行：macOS / Windows / Linux 全系统适配
✅ Recursive Deep Scan: Scan all images in root and subfolders ✅ Custom Similarity: 0~100% adjustable threshold ✅ Safe Soft-Delete: Only move files, never delete originals ✅ Smart Selection: Keep high-resolution & large-size images ✅ Full Format Support: JPG/PNG/WebP/BMP/TIFF ✅ Cross-Platform: Compatible with macOS/Windows/Linux
使用方法 | Usage
AI 智能调用 | AI Agent
中文：帮我扫描 [文件夹路径] 下所有图片，使用 [相似度] 进行去重 English: Scan all images in [folder path] with [similarity] deduplication
命令行调用 | CLI
# 中文示例
python3 {baseDir}/main.py --path "你的图片目录" --similarity 99.5
# English Example
python3 {baseDir}/main.py --path "Your Image Folder" --similarity 99.5
Usage Guidance
This skill appears to do what it claims: recursively find duplicate images and move lower-quality copies into a local '重复图片回收站' folder under the path you provide. Before running: (1) back up important photos (the tool moves files, it does not permanently delete them), (2) ensure Python 3 and Pillow (pillow>=10) are installed in the environment, (3) test on a small folder to validate similarity thresholds and behavior (default similarity 99.5), and (4) note the script ignores files smaller than ~10KB and skips the trash folder by name. No network access or credential use was found. If you need stricter guarantees, inspect and run the included main.py locally in a safe test directory first.
Capability Analysis
Type: OpenClaw Skill
Name: deduplicate-images
Version: 1.0.1
The skill bundle is a legitimate image deduplication tool that performs recursive directory scanning and moves duplicate images to a local 'trash' folder. The code in main.py uses standard libraries (Pillow, shutil) for image processing and file management, and its behavior aligns perfectly with the documentation in SKILL.md and README.md without any signs of data exfiltration, malicious execution, or prompt injection.
Capability Assessment
Purpose & Capability
Name/description (image deduplication, safe soft-delete, keep high-quality originals) align with the delivered code and runtime instructions. The script scans directories, computes simple image fingerprints, compares similarity, and moves lower-quality duplicates into a local '重复图片回收站' under the target folder — all consistent with purpose.
Instruction Scope
SKILL.md and main.py limit actions to scanning image files in the provided path, skipping the trash folder, and moving files to a local trash subfolder. No instructions or code read unrelated system files, environment variables, network endpoints, or other skills' configs. The tool does perform file moves (not permanent deletion), which is expected behavior but is a destructive file-operation permission the user should authorize.
Install Mechanism
There is no install spec (instruction-only), which keeps risk low. SKILL.md metadata lists a pip dependency ('pillow>=10.0.0') but the registry's top-level requirements did not explicitly list pip packages; users must ensure Pillow is installed in their environment. No downloads or external installers are performed by the skill.
Credentials
The skill requests no environment variables or external credentials. The only runtime requirement is python3 and Pillow (PIL). It operates on the user-specified filesystem path only, which is proportionate to its function.
Persistence & Privilege
The skill is not marked always:true and does not request persistent system-wide changes. It creates a trash folder inside the provided root folder and moves files there (normal for this tool). It does not modify other skills or global agent settings.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install deduplicate-images - After installation, invoke the skill by name or use
/deduplicate-images - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
Image Deduplicator Pro 2.0.0
- Added recursive scan for all images in root and subfolders.
- Introduced customizable similarity threshold (0-100%).
- Safe deduplication by only moving duplicates to trash, originals are preserved.
- Automatically keeps high-resolution and large-size images.
- Expanded format support: JPG, PNG, WebP, BMP, TIFF.
- Now cross-platform: works on macOS, Windows, and Linux.
v1.0.0
- Added initial release of deduplicate_images skill.
- Recursively scans root and all subfolders for image duplicates.
- Lets users set a custom similarity threshold for deduplication.
- Moves lower quality duplicates to a centralized recycle bin for safe deletion (no files are deleted).
- Prioritizes keeping images with higher resolution and larger file size.
- Automatically excludes the recycle bin from scans to prevent loops.
- Supports multi-threading, all image formats, and handles name conflicts safely.
Metadata
Frequently Asked Questions
What is 图片智能去重 - 递归全目录·安全回收站版?
递归扫描全目录图片，自定义相似度，安全软删除至回收站，保留高清原图. It is an AI Agent Skill for Claude Code / OpenClaw, with 139 downloads so far.
How do I install 图片智能去重 - 递归全目录·安全回收站版?
Run "/install deduplicate-images" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is 图片智能去重 - 递归全目录·安全回收站版 free?
Yes, 图片智能去重 - 递归全目录·安全回收站版 is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does 图片智能去重 - 递归全目录·安全回收站版 support?
图片智能去重 - 递归全目录·安全回收站版 is cross-platform and runs anywhere OpenClaw / Claude Code is available (darwin, linux, win32).
Who created 图片智能去重 - 递归全目录·安全回收站版?
It is built and maintained by onealmeng (@onealmeng); the current version is v1.0.1.
More Skills