Changelog
What shipped, when, and what we broke on the way. Incidents are listed here too — not on a separate page where nobody looks.
Where does Klip publish incidents?
Klip publishes incidents in this changelog, in the same list as features, with the same dates and the same plain language. There is no separate status page to check: if a deploy broke something, it appears here with what was affected, what was not, and what changed to stop it recurring.
A code that has been turned off, expired, mistyped or hit during an outage now gets a page that says which of those happened and whose fault it is. They carry no JavaScript and no web font: a page explaining a failure should not be able to fail twice.
A scan comes from a camera, a click comes from a tap, and Slack fetching a preview is neither. All three are now measured apart, and the chat-app fetches never join the total. Expect the headline number to be smaller than a naive counter would report, and correct.
The QR endpoint refuses a foreground and background pair below a 3:1 luminance ratio and says why, rather than rendering a code that looks fine on a monitor and fails on paper in a dim room.
Disabling a link writes a blocklist entry read on every redirect, which the platform caches for a minimum of 30 seconds. So the real worldwide propagation is about half a minute rather than instant. We had documented it as near-immediate; that was wrong and it is now stated correctly.
Both work signed out, both render through the same engine that produces printed files, and both include the four-module quiet zone. The size calculator uses the 10:1 rule and shows the practical print floor separately.
No incidents yet — nothing is deployed to write one about. When there is one, it goes here with the window, what kept working, and what changed afterwards.