Skip to content
checkaifree

Local tools

Free image metadata tools

Five browser tools for the metadata around an image: EXIF and captions, Content Credentials, a cleaned download, Stable Diffusion PNG text, and a ComfyUI workflow. Each file stays on your device. None of these pages verifies a signature or runs a workflow.

17 October 2026 · Check AI Free

What these pages are for

People looking at a suspicious picture usually need one concrete file operation before they need a score. They want the GPS block, or they want it gone. They want to know whether a Content Credentials manifest is still in the JPEG. They have an original PNG and they want the ComfyUI graph that was saved inside it. Those are different jobs from “is this image AI?”, and they deserve their own page, their own heading, and a tool that does only that job. The image checker remains the place for source clues and an optional model check. It can send file-header data and a smaller image copy when a pixel reading is required. The tools on this index do not.

The split also matches the way the containers work. EXIF, XMP and IPTC are camera and caption directories. A C2PA manifest is a claim, a list of actions and a signature this site does not verify. A ComfyUI workflow is JSON in a PNG text chunk, which ComfyUI’s own docs say is not a signature. Stripping metadata is a rewrite of those containers, either by copying the compressed pixels without the metadata segments or, when the container cannot be rewritten, by letting the browser encode a new file. Mixing those into one unlabeled drop zone makes it easy to think a missing field, a manifest and a model score are the same kind of evidence. They are not.

Use the original file when you still have it. A screenshot is a new picture of a screen, and a messenger download is often a new compression. The metadata survival report is edition 1 of a small local experiment on that point: camera fields in eight photographic originals were absent from a specified JPEG export and from a browser screenshot. It is not a ranking of social networks. The sharing guide says what that limit means when the picture came from an app. The Content Credentials guide says what to do when you need a validator rather than the reader on this site.

If you are integrating the checker rather than inspecting one file, the developers page describes the HTTP endpoints and the checkaifree-mcp package that actually exist. Those paths can upload a selected image. The tools in this list do not. Read the method page before you treat any of it as proof.

Questions

Do these tools upload the image?

No. Each tool reads or rewrites the file in your browser. The image checker linked from these pages is separate: that checker can send a smaller copy to a detection provider when a pixel check is needed.

Do they prove an image is real or AI-generated?

No. They show what a container still holds. A camera name, a manifest or a workflow is a stored claim. It can be missing, edited or copied. The image checker is the separate tool for a source-clue and optional model reading.

Why are the tools separate from the detector?

Viewing EXIF, reading a manifest and extracting a workflow are specific jobs with their own search. Keeping them as their own pages makes the limit of each one obvious: this file, these fields, no signature check, no workflow execution.

Related