image-metadata-cleaner

Clean privacy-sensitive metadata (C2PA, EXIF, XMP, IPTC, GPS) from user-owned images by writing sanitized copies. For privacy hygiene and file preparation.

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "image-metadata-cleaner" with this command: npx skills add aiwork4me/image-metadata-cleaner

Image Metadata Cleaner

Clean privacy-sensitive metadata from user-owned images by writing sanitized copies. Designed for legitimate privacy hygiene, file preparation, and reproducible publishing workflows.

Use only for: images you own or are authorized to process, for privacy protection, file-size optimization, and clean publishing.

What it does

  • Re-encodes image pixels into a fresh output file — all metadata discarded
  • Writes copies instead of modifying originals in place
  • Defaults folder output to metadata-cleaned/ subdirectory
  • Refuses output paths that resolve to the same file as the input
  • Reopens outputs and scans for residual metadata keys and provenance markers
  • Produces a human-readable summary and optional JSON manifest

Usage

Single file

User: clean metadata from this image: photo.png
User: remove EXIF data from IMG_2024.jpg

Folder batch

User: clean metadata from all images in "C:\Users\me\Downloads"
User: remove privacy data from /path/to/folder

Steps

  1. Confirm the task is for privacy hygiene on images the user owns or is authorized to process.

  2. Preview with dry-run (optional):

    uv run --with "pillow>=10.0" scripts/strip.py "<path>" --dry-run
    
  3. Run the cleanup:

    uv run --with "pillow>=10.0" scripts/strip.py "<path>" --manifest
    

    Options:

    • -o <path> — Output file path (single file only)
    • --output-dir <dir> — Output directory (batch mode)
    • -f preserve|jpg|png — Output format (default: preserve — JPEG stays JPEG, others become PNG)
    • -q <1-100> — JPEG quality (default: 95)
    • --recursive — Process subdirectories
    • --overwrite — Overwrite existing output (only after user confirmation)
    • --dry-run — Preview without writing files
    • --manifest [path] — Write JSON manifest

    If uv is not available:

    pip install "pillow>=10.0" && python scripts/strip.py "<path>" --manifest
    
  4. Report results:

    • Files processed, failed, or dry-run previewed
    • Output filenames and location
    • File size before → after
    • Dimensions preserved
    • Verification scan results (any residual metadata keys or provenance markers)

    Note: this removes file-level metadata only. Pixel-level watermarks and external platform records are outside the scope.

Error Handling

ErrorCauseFix
"No supported image files found"Folder has no matching extensionsCheck input path and file types
"Output already exists"Explicit output path existsUse --overwrite after user confirmation
"Refusing to overwrite input file"Output path = input pathChoose a different output path
"Unsupported image extension"File extension not in supported listUse PNG, JPEG, WebP, BMP, or TIFF
"cannot identify image file"Corrupted or non-image fileSkip and continue with other files
Pillow ImportErrorMissing dependencyRun pip install "pillow>=10.0"

What gets removed

Metadata TypeNotes
EXIFCamera, GPS, device tags. Orientation applied before saving.
XMPAdobe and application metadata
IPTC/PhotoshopPress and photo metadata
ICC ProfileColor profile (not copied to output)
C2PA/JUMBFProvenance containers removed by re-encoding

Output behavior

  • Format: Default preserve — JPEG inputs stay JPEG, others written as PNG
  • Naming: {name}-clean.{ext} (e.g., photo-clean.png)
  • Folder mode: Outputs go to metadata-cleaned/ subdirectory
  • Single file: Sibling copy next to the original
  • Never overwrites input — script refuses if output resolves to input

Supported inputs

.png, .jpg, .jpeg, .webp, .bmp, .tiff, .tif

Known limitations

  • Does not remove pixel-level watermarks, fingerprints, or invisible signals
  • Does not affect external platform records or server-side provenance
  • Removing ICC profiles may affect color management in some workflows
  • JPEG output is lossy; PNG preferred for pixel fidelity
  • Transparent images written as JPEG are composited onto white background

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

Miaoji Bid Guard

亚马逊广告出价护栏顾问。当用户需要设定ACoS上限、分析广告关键词出价、制定熔断策略、生成视频广告脚本时使用。支持:利润边界设定、秒级出价方案生成、熔断建议、视频广告内容创作。

Registry SourceRecently Updated
General

Miaoji Scene Studio

快速生成适配TikTok等平台的产品场景图AI绘图提示词与构图方案,提升内容真实感与点击率。

Registry SourceRecently Updated
General

Miaoji Model Shot

基于服装描述快速生成专业模特拍摄构图方案和AI绘图提示词,高效规划新品展示方案与拍摄指导。

Registry SourceRecently Updated
General

Miaoji Compliance Copy

快速扫描并清洗亚马逊产品文案违规词,注入场景和用户意图,生成符合平台规则的多版本合规文案。

Registry SourceRecently Updated