Metadata Cleaning API

At a glance

Base address https://metadatatools.io. One file per request. The clean file is the response body and the outcome travels in headers, so there is nothing to parse.

Photos and PDFs

Post the file as multipart/form-data in a field named file, or as the raw body.

# photo (JPEG, PNG, WebP)
curl -X POST https://metadatatools.io/api/v1/clean/image \
  -H "Authorization: Bearer $KEY" \
  -F "[email protected]" -o holiday-clean.jpg

# PDF
curl -X POST https://metadatatools.io/api/v1/clean/pdf \
  -H "Authorization: Bearer $KEY" \
  -F "[email protected]" -o invoice-clean.pdf

// Node.js 18 or newer
import fs from "node:fs";

const form = new FormData();
form.append("file", new Blob([fs.readFileSync("holiday.jpg")]), "holiday.jpg");

const res = await fetch("https://metadatatools.io/api/v1/clean/image", {
  method: "POST",
  headers: { Authorization: `Bearer ${process.env.KEY}` },
  body: form,
});
if (!res.ok) throw new Error(await res.text());
console.log("removed", res.headers.get("x-metadata-removed"), "left", res.headers.get("x-metadata-remaining"));
fs.writeFileSync("holiday-clean.jpg", Buffer.from(await res.arrayBuffer()));

# Python
import os, requests

with open("holiday.jpg", "rb") as f:
    r = requests.post(
        "https://metadatatools.io/api/v1/clean/image",
        headers={"Authorization": "Bearer " + os.environ["KEY"]},
        files={"file": f},
    )
r.raise_for_status()
print(r.headers["X-Metadata-Removed"], "removed;", r.headers["X-Metadata-Remaining"], "left")
open("holiday-clean.jpg", "wb").write(r.content)

Pixels are copied, not re-compressed, so quality is exactly what you sent.

Video and RAW

These files are large, so send the bytes as the request body and put the file name in the X-Filename header. The call stays open while our server works and then returns the clean file. Video is remuxed without re-encoding, so picture and sound are untouched. Each file spends one of your monthly video or RAW cleanings.

# video: .mp4 .m4v .mov .mkv .webm
curl -X POST https://metadatatools.io/api/v1/clean/video \
  -H "Authorization: Bearer $KEY" -H "X-Filename: holiday.mov" \
  --data-binary @holiday.mov -o holiday-clean.mov

# RAW: .cr2 .cr3 .nef .arw .dng .raf .orf .rw2 and more
curl -X POST https://metadatatools.io/api/v1/clean/raw \
  -H "Authorization: Bearer $KEY" -H "X-Filename: DSC_1024.nef" \
  --data-binary @DSC_1024.nef -o DSC_1024-clean.nef

For RAW, location, capture time, owner name, serial numbers and XMP and IPTC blocks are removed, while the image data and the tags a RAW editor depends on stay put. Nikon uses its camera serial number as a key to decode its own data, so it is kept and appears in X-Metadata-Left along with anything else that could not be taken out. Set a generous client timeout for big files.

Look before you clean

POST /api/v1/inspect takes any file as the raw body and returns every tag it holds as JSON, without changing the file. It uses one image-and-PDF allowance, like a clean call, and a failed read gives it back.

curl -X POST https://metadatatools.io/api/v1/inspect \
  -H "Authorization: Bearer $KEY" \
  --data-binary @holiday.jpg

Reading the response

Every clean call reports its outcome in headers:

X-Metadata-Removed

Number of metadata fields found and deleted.

X-Metadata-Remaining

Fields still present after a second scan. 0 means verified clean.

X-Metadata-Left

Names of fields that could not be removed, URL-encoded and comma-separated. Empty for a clean file.

X-Metadata-Fields

Names of up to 30 removed fields, URL-encoded and comma-separated.

X-Quota-Limit, X-Quota-Remaining, X-Quota-Period

What is left of the allowance this call drew on: images and PDFs per day, video or RAW per month. Unlimited plans report unlimited.

Treat a file as clean only when X-Metadata-Remaining is 0. To see your balance without spending anything, call GET /api/v1/usage: it returns your plan, what you used and have left of each allowance (a limit of null is unlimited) and the largest file you may send.

Prefer to talk to an assistant instead of writing code? The MCP server wraps this API.

Allowances per plan

The daily image-and-PDF count is shared by the website, the API and MCP and resets at midnight UTC. Video and RAW counts reset every month. Requests are also limited to 120 a minute per account. More room? Compare plans

When a call fails

Failures come back as JSON in this shape: { "error": { "code": "…", "message": "…" } }

On a 429 with a Retry-After header, wait that long. 413, 415 and 422 will not succeed on a retry; log them and move on. Bulk recipes are in cleaning files in bulk.