Hand an AI agent a video URL and it does what a person would have to do, which is scrape the page. It burns context reading markup built for human eyes and still cannot tell you what happened at a given moment in the recording.
How to do it
- 1
Copy the share link for any recording
- 2
Add .json to the end of it
So quicksnip.com/s/abc123 becomes quicksnip.com/s/abc123.json
- 3
Hand that URL to your agent instead of the share link
The response tells the agent what it has and what to fetch next, so it does not have to guess.
What you end up with
A structured document with the title, an ISO timestamp for when it was recorded, the duration, the dimensions, any chapters, and a set of still frames each with its own millisecond offset and its own fetchable URL.
Questions
Does this work on screenshots too?
Screenshots have their own shortcut. Add a plus sign to the share link for the raw image, which is what you want when an agent or an issue tracker needs the file rather than the page.
Do I need to enable anything first?
No. Every share link already answers on .json.
Try it yourself
QuickSnip is free to start. Record or snap, and the link is on your clipboard before the window closes.
Download QuickSnipNext
Put a screenshot straight into a GitHub issue
Add a plus sign to the share link and you get the raw image, which is what an issue tracker needs to render it inline.
Record a bug report a developer can actually fix
Narrate what you were doing while you do it, so the recording carries the steps instead of just the symptom.