How to Remove Metadata From Hundreds of Files: Web App, API or MCP

One photo is a click. Three hundred is a plan. There are three ways to clean in bulk here, and three questions choose between them: how many files, how often, and how sensitive.

Quick comparison

Browser: when confidentiality matters most

Drop the files, press Clean, download one .zip. Images and PDFs stay on your device. Daily allowance for those: 50 as a guest, 500 on a free account, unlimited on paid plans. A few dozen files is comfortable. Start with the image or PDF tool.

API: when the job repeats

Each request sends one file and returns the clean one, with results in headers. Endpoints: /api/v1/clean/image, /pdf, /video, /raw (see the API docs). Allowances match the site: 500 images and PDFs a day on a free account, no daily cap on paid plans, monthly video and RAW quotas everywhere.

for f in *.jpg; do
  curl -s -X POST https://metadatatools.io/api/v1/clean/image \
    -H "Authorization: Bearer $KEY" \
    -F "file=@$f" -o "clean-$f" -D headers.txt
  grep -i "x-metadata-remaining" headers.txt
done

import fs from "node:fs";

const KEY = process.env.KEY;
const files = fs.readdirSync(".").filter((f) => /\.(jpe?g|png|webp)$/i.test(f));

for (const name of files) {
  const form = new FormData();
  form.append("file", new Blob([fs.readFileSync(name)]), name);
  const res = await fetch("https://metadatatools.io/api/v1/clean/image", {
    method: "POST",
    headers: { Authorization: `Bearer ${KEY}` },
    body: form,
  });
  if (!res.ok) { console.error(name, res.status, await res.text()); continue; }
  console.log(name, "removed", res.headers.get("x-metadata-removed"), "left", res.headers.get("x-metadata-remaining"));
  fs.writeFileSync(`clean-${name}`, Buffer.from(await res.arrayBuffer()));
}

Video and RAW requests wait while the file is processed, so allow a long timeout. Write output to another folder or add a suffix so originals are never overwritten.

AI assistant: when describing the job is quicker than scripting it

The MCP server, included on every plan, lets you say “clean everything in ~/Photos/shoot and tell me what was in it.” A request covers up to 200 files, two at a time. Each copy is saved beside its original as name-clean.ext and every file is reported. Folders are not searched recursively and existing clean copies are skipped, so repeating a request is harmless. Setup: removing metadata with Claude, Cursor or ChatGPT.

Which one?

• Under about 30 images or PDFs, or anything sensitive: browser.
• Hundreds of files, or recurring: API.
• One folder, already working in Claude or Cursor: assistant.
• Video or RAW in bulk: API or assistant; mind monthly quotas and per-file size.

Do the sums: 300 photos equal 300 images, which one day on a free account (500 a day) covers. If they are RAW, each spends a monthly RAW cleaning instead: 10 on Free, 100 on Starter, 1,000 on Pro, wherever you start, because RAW is always processed on our server.

Good habits for any bulk run

1. Clean copies; never make a clean file your only original.
2. Count files in and out and scan the log for anything not verified clean.
3. Spot-check one with exiftool -a -G1 -s file or the metadata viewer.
4. Keep originals privately; they record where and when.
5. Rename on purpose, as file names carry places and dates.

What is removed per format: images, PDFs, video, RAW.