You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
ElsewhereArea: what it touches
ImprovementsKind: in v2.1.292,
ImprovementsSection of the release
What
In a session connected through Remote Control, messages that wait while earlier ones are still being sent now go into a capped queue. If more than a set number of updates from subagents (helper agents Claude starts to handle part of a task) pile up, the oldest are dropped until the queue empties.
If the cap itself fails, the failure is caught and written to the log.
Why
In Remote Control sessions with a lot of subagent activity, older subagent updates can now be lost without any notice, for example after a reconnect.
