You'll noticeTier: how much it should matter to you
1Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
ElsewhereArea: what it touches
ImprovementsKind: in v2.1.323,
ImprovementsSection of the release
What
Claude Code can send an extra background request to keep the conversation's cache warm, so the next reply can reuse work already done. It now skips that request in more situations:
the last turn was handed off
the chat holds uploaded files
the chat holds pictures returned by tools
there has been no reply since compaction, which is when Claude Code summarises the conversation to save space
the last reply has no model recorded
the last reply came from a different model
a new memory-related check says not to
Each skip is logged with its reason, and no warming request is sent.
Why
Fewer warming requests are wasted on a cache that would not match the next reply, such as after compaction or after you switch models.
