First public release of the VideoContext Chrome extension — capture a portion of your screen, extract the most useful frames, and paste them as a single grid into Claude, ChatGPT, or Gemini.
Video isn't a supported input in any major LLM today. VideoContext bridges that gap: record, pick the frames that matter, copy a stitched PNG to the clipboard, paste.
What's in this release
Capture
- Tab / Window / Screen capture via Chrome's native picker
- Optional area-select — draw a rectangle to crop every frame
- 3-2-1 countdown between source pick and recording, so you can set the stage
- Recording strip with Stop / Pause / Discard while capture is in progress
Presets — all target 10 frames
- Quick — 5s at 2 fps. UI demos, error screens, fast flows.
- Standard — 10s at 1 fps. Walkthroughs, multi-step tasks.
- Slow — 20s at 0.5 fps. Long processes, slow-moving content.
Delivery
- Copy as Grid — selected frames stitched into a single watermarked PNG and written to the clipboard. One paste, all frames.
- Save PNGs (zip) — bulk export as
frame-001.png,frame-002.png, … inside a single archive.
Privacy by design
- 100% offline. No accounts, no servers, no telemetry.
- Frames live in browser memory only — closing the widget discards them.
- Uses
activeTabinstead of a broad<all_urls>host permission.
Install
Chrome Web Store
Listing pending — link will be added here once the review is approved.
From source (unpacked)
- Download the Source code (zip) below and unzip it.
- Open
chrome://extensionsin Chrome. - Toggle Developer mode on (top right).
- Click Load unpacked and select the unzipped folder.
- Pin the VideoContext icon to your toolbar.
Try it on real bugs
Five live, intentionally buggy demos to record on: videocontext.app/examples.html
License
MIT.
Built by @singh_jasminder. If it saves you time, you can chip in toward the next round of features.