← all repositories
Inginnng/EditHere

Stop describing pixels in prose — just drag them

EditHere turns visual feedback into structured JSON with pixel coordinates, so any multimodal AI knows exactly what to change.

EditHere
Collecting fresh signals — velocity needs a few days of history.
collecting data…
star history

What it does EditHere is a local-first C++ annotation and layout tool for giving AI visual feedback. You screenshot (or open) any image or video, click or box regions, write comments, or literally drag UI elements to new positions — and it exports a clean original image plus a JSON file with numbered annotations and before/after coordinates. It works on anything you can see on screen: web pages, native apps, games, AI-generated images, even paused video frames.

The interesting bit The “Big Bang” mode segments the screen into draggable rectangles, so instead of typing “move that button left a bit,” you just move it — and the tool records the exact source and destination regions for the AI to act on. It deliberately avoids DOM-based annotation, which is why it works on games and desktop software where browser-side tools are blind.

Key highlights

  • Annotations stay out of the image: the AI gets a pristine original plus separate coordinates, not red boxes burned into pixels
  • Video annotation with timestamps and frame captures — only frames you actually commented on get exported
  • Built-in OCR using system engines (Windows OCR / macOS Vision), fully offline, no extra dependencies
  • Long-scroll capture, color picking, magnifier with pixel coordinates, and pinned screenshots
  • Ships an edithere-cli and a skill so agents like Claude Code or Codex can open the editor and wait for your feedback

Caveats

  • PolyForm Noncommercial 1.0.0 license — free for personal use, but commercial use is off the table
  • The README is Chinese-first (an English version exists); the UI supports both languages
  • Newer versions can read old feedback files, but old versions can’t read v4 — worth knowing if you batch-update

Verdict If you iterate on UIs, game HUDs, or AI-generated images with an AI assistant and are tired of typing “the button near the top right, a bit lower,” this solves exactly that. If you never collaborate with a multimodal model, it’s a competent screenshot tool with a niche export format you won’t use.

Frequently asked

What is Inginnng/EditHere?
EditHere turns visual feedback into structured JSON with pixel coordinates, so any multimodal AI knows exactly what to change.
Is EditHere open source?
Yes — Inginnng/EditHere is an open-source project tracked on heatdrop.
What language is EditHere written in?
Inginnng/EditHere is primarily written in C++.
How popular is EditHere?
Inginnng/EditHere has 1k stars on GitHub.
Where can I find EditHere?
Inginnng/EditHere is on GitHub at https://github.com/Inginnng/EditHere.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.