Red Annotation Extractor
Ask an AI image model to mark up a photo and it returns a subtly re-rendered image rather than your original. This isolates just the red annotations as a transparent PNG - with a sensitivity slider - and can composite them back onto your untouched original, so you get clean marks over pristine pixels. Runs fully client-side; neither image is uploaded. Imported Claude artifact. Source: https://github.com/simonw/tools (red-extractor.html, Apache-2.0).
Every public app is built from a security-scanned skill and must pass a clean scan — skill and frontend — before it can be listed. Have a skill of your own? Turn it into an app — or read the step-by-step walkthrough.
Name your fork
This clones the app's prompt and source into your account as a new unlisted app, then opens it in the AI developer so you can make it your own.
We'll put this in the AI developer's message box for you to review and send.
View source prompt + release files — fork-enabled apps share their full source
System prompt (SKILL.md)
---
name: red-annotation-extractor
description: Pull the red markings out of an AI-annotated image as a transparent PNG, then overlay them back onto the untouched original.
metadata:
tags:
- artifact-import
- media
source_url: https://tools.simonwillison.net/red-extractor
github_repo_url: https://github.com/simonw/tools
---
# Red Annotation Extractor
An imported Claude artifact, built around a specific annoyance with AI image
models: ask one to mark up a photo with red circles or arrows and it hands back
a subtly re-rendered image, not your original with marks on it.
The fix is two steps. Upload the annotated image and a sensitivity slider
isolates just the red pixels, everything else transparent - downloadable as a
PNG on its own. Then optionally upload the untouched original and the extracted
annotations composite on top, so you keep clean marks over pristine pixels.
Pixel work runs on a canvas in the page. Neither image is uploaded anywhere, no
model is called, and nothing is metered.