VideoContext App

About

VideoContext App turns screen recordings into something you can actually paste into an LLM.

Video isn't a supported input in Claude, ChatGPT, or Gemini today. VideoContext records your screen, extracts the most useful frames as images, and stitches them into a single grid you can copy with one click and paste straight into a chat.

▸ How it works
  1. Click the toolbar icon — a floating recorder appears on the page.
  2. Pick Tab, Window, or Entire Screen, choose a duration, optionally drag a region to crop.
  3. Hit Record. A 3-2-1 countdown gives you a moment to set the stage, then VideoContext silently captures frames at a steady cadence.
  4. When it stops, review the frame grid in the widget. Select the ones you want. Copy as Grid → paste into your LLM.

▸ Built for LLMs
  • Default preset captures 10 frames — within ChatGPT's and Gemini's per-message limits and comfortable for Claude.
  • Copy as Grid stitches selected frames into a single PNG so one paste sends the whole sequence.
  • Save PNGs (zip) downloads the frames as a single archive if you want to keep them.

▸ Three duration presets, plus custom
  • Quick — 5s, 2 fps → 10 frames. UI demos, error screens, fast flows.
  • Standard — 10s, 1 fps → 10 frames. Walkthroughs, multi-step tasks.
  • Slow — 20s, 0.5 fps → 10 frames. Long processes, slow content.

▸ Privacy by design
  • 100% offline. No accounts, no servers, no telemetry.
  • Frames are held in browser memory during a session and dropped when you close the recorder.
  • The extension uses the activeTab permission — it only sees the tab you're on when you click the icon. It can't read any other tab.

▸ Coming soon
  A native macOS menu-bar app for system-wide capture with global hotkeys.

Find us at videocontext.app