remove_metadata Leave feedback

remove_metadata strips metadata and saves a cleaned copy. With no categories it removes everything removable — the usual “clean this before I send it”. Pass categories to remove only certain kinds. Example prompt: “Strip all metadata from these files before I share them”.

Tool description (as the AI agent sees it):

Removes metadata from a document and saves a cleaned copy to storage. By default (no categories) it strips ALL removable metadata — use this for ‘remove all metadata before sharing’. To remove only SPECIFIC kinds, pass categories (one or more of): gps (location/geotags), author (creator/editor), comments, company, dates, software (the ‘created with’ tool fingerprint), copyright, keywords, personal (best-effort PII bundle: people + company + location). Call this whenever the user asks to remove, strip, clean, or redact metadata — e.g. ‘remove GPS from photo.jpg’ or ‘strip the author from report.pdf’. Supports PDF, DOCX, XLSX, PPTX, JPEG, PNG, TIFF and 100+ more formats. Returns a saved-path message with the number of properties removed. On failure the response text starts with ‘Metadata removal failed for’.

{
  "name": "remove_metadata",
  "arguments": {
    "file": {
      "filePath": "report.pdf"
    }
  }
}

A saved-path message naming the cleaned file. The original is untouched — which is the point, and also the trap: share the cleaned copy, not the original.

What “everything” means is bounded by what the engine can remove: document properties, EXIF/XMP/IPTC packages, and similar. Information that is structurally part of the document — tracked changes, embedded objects, earlier revisions — is not metadata and is not removed here.

On failure the text starts with Metadata removal failed for, followed by the exception type and message.

• “Strip all metadata from this file before I share it.”
• “Remove just the EXIF data from these photos.”
• “Clean the author and company fields but keep the dates.”

See it used end-to-end: Strip metadata before sharing.