Follow Discord
Sweep 10 Nov 2026 · 02:18Z Build v2.1.322 539 read Stable v2.1.311 Latest v2.1.322 Next v2.1.322 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.271 ·

Reminders can now survive conversation compaction and thread continuation

Batching and secondary reminders now survive transcript compaction instead of being dropped or needlessly resent

Group of 5 You'll notice Improvements
JSON All of v2.1.271
You'll noticeTier: how much it should matter to you
3Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
CompactionArea: what it touches
ImprovementsKind: in v2.1.271,
ImprovementsSection of the release

What

  • The transcript-folding (compaction) pipeline now keeps a record of which reminders (batching_reminder_sent, secondary_reminder_sent) have already been sent, and only re-sends ones whose text actually changed, via a reminderFromRecords flag.
  • A new diffing helper computes which reminders need to be resent by comparing freshly computed reminder text against the last-sent record.
  • The server-side "tether" logic that decides whether a thread can continue instead of starting fresh can now tolerate a transcript that ends in an ephemeral system reminder, when a new parameter is set; previously any trailing ephemeral reminder forced a new thread. It also records keptReminderClearAt and keptReminderScope in telemetry.
  • Message normalization gained a keptReminders flag, true when the caller requests it and the model response includes the beta mid-conversation-system-clear-at-2026-08-21.
  • A matching mid_conversation_system_clear_at beta header was added to the list of recognized Anthropic API betas.

Why

Together these changes mean reminder messages (like batching or secondary reminders) survive conversation compaction and thread continuation instead of being dropped or needlessly resent, keeping the model's context consistent across long sessions.

See this entry in the whole of v2.1.271 →

Feedback