Tips, tutorials, and thoughts on making videos people actually watch.
A practical design-QA loop: point with Flare, describe a measurable delta, let the developer’s agent patch it, then review recorded proof instead of another comment thread.
TL;DR * Neither macOS nor Windows can screenshot a full scrolling page natively — their tools stop at the visible screen. * Chrome can
TL;DR * AI agents finish tasks and report "done." A diff proves code changed — it does not prove the app
TL;DR * Clipy's screenshot editor runs in your browser at clipy.online/screenshots — free, no watermark, no install needed for
TL;DR * Screenshot attachments die in chat scrollback. A link is permanent, always current, and opens for anyone — no account, no app,
There is now a free plan and Clipy Pro, and nothing in between. The reasoning, what stays free, what Pro adds, and what to do if the price is wrong for you.
Paste a URL and you get title-guessing. Three escalating methods for giving Claude Code, Codex or Cursor the actual contents of a video — transcript, gaps and frames.
Not the support-desk kind. A private, searchable library of what you actually watched — that keeps the visual half and that your coding agent can read mid-task.
Public videos only, captions required, transcript text only, and sources that vanish in 30 days. Every limit cited to Google, and the one worth switching tools over.
The fastest way to verify an AI agent's fix is not reading the diff. It is watching the work. Make your agent record its own proof with Clipy, and let another agent confirm it.
AI agent screen recording now runs in both directions. Agents could already read a Clipy recording as text and frames. With the
To record your website at different screen sizes, you normally resize a browser window three times and stitch three clips together. The