Repository navigation
v1.4.2
Immutable
release. Only release title and notes can be modified.
caret 1.4.2 adds a band above Claude Code's prompt while caret holds a plan for review, with a link that opens that review. It also smooths the review surface: plan swaps crossfade instead of jumping, every state transition runs at one 160ms pace, and caret doctor lists failing checks first.
Added
- While caret holds a plan for review, Claude Code shows a 🥕 band above its prompt, so the session no longer looks idle. Its link, or
1in an empty prompt, opens that session's review in the browser, already logged in when token auth is on. Approving, denying, cancelling or timing out clears it. Turn it off under Settings → Notifications or withreview.band = false; the change applies from the next review. The band needs Claude Code 2.1.287 or later; older versions keep caret's hooks and show no band. caret doctorandcaret installtake--no-color, and both print plain text underTERM=dumbas they already did underNO_COLORandCI=true.
Changed
- Every plan swap crossfades: a plan arriving, the next plan after a decision, switching plans, opening one from a notification, and the queue emptying. A plan renders highlighted on its first paint instead of recolouring a moment later, and Approve and Request Changes stay busy until the next plan is on screen.
- Menus, dialogs, popovers and alerts open and close at one 160ms ease-out instead of three different durations.
- Key hints are hidden on touch devices, whatever the shortcut-hints setting says.
caret doctoropens with a header line, then the failing checks in one block with each fix beside it, and ends on the next command to run.- caret's sound cues use cuelume's new palette, so every cue sounds different.
Fixed
- In the caret-light theme, faint text and the bright accent meet 4.5:1 contrast on every surface, which also makes the primary button's hover label readable.
- A plan table whose delimiter row is missing a cell renders as a table instead of raw pipes.
Full Changelog: v1.4.1...v1.4.2