You'll noticeTier: how much it should matter to you
5Useful: my rating, 1 to 5
5Signal: worth watching, 1 to 5
Prompt CacheArea: what it touches
In DevelopmentKind: in v2.1.286,
What probably matters to youSection of the release
What
The prompt cache lets the API reuse the start of a conversation instead of processing it again. A new module keeps a one-hour cache entry alive by sending a small request to /v1/messages/cache_touch at a regular interval.
Settings come from the server-controlled switch tengu_sequential_cloud, which holds touchIntervalSeconds (10 to 120, default 30) and activityWindowMinutes (1 to 1440, default 60).
It runs only for subscribers with a one-hour cache lifetime. It does not run in a cloud session, when using overage (usage beyond your plan), or when ANTHROPIC_UNIX_SOCKET is set.
The requests carry the header anthropic-cache-keepalive.
A new policy entry, allow_cache_keepalive, labelled "Renewing the prompt cache lifetime" (requirement HIPAA-R65), must also allow it. It is denied for hipaa organizations and is denied when the policy cannot be read.
Why
Keeping the cache alive could mean fewer cache misses after idle pauses in long sessions. Organizations under HIPAA rules will not get it. Nothing has been read about tengu_sequential_cloud, so whether keepalive is running is not known.
Read from
Feature flag
tengu_sequential_cloudNot enough to say
Nothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.
This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.286: off
These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.
