A new staleServerClassifierContext check re-verifies auto-mode artifact permission approvals before they're used, retrying if facts changed
What
A new staleServerClassifierContext() method was added to the artifact permission logic. When a server-side classifier approves an action in auto mode (where permission decisions are made automatically rather than asked of the user), this check now looks at whether key facts about the artifact have changed since the request was built, such as ownership, audience, whether it's being shared live, whether it was co-written, or whether it was already attested/watched.
If those facts changed, the previously-granted allow is voided and the request is blocked so it can be retried, and a new tengu_auto_mode_server_context_stale event is logged.
Why
This prevents an auto-mode approval from being used based on outdated information about an artifact, so a permission decision that was correct when first made can't be reused once the underlying situation has actually changed.
tengu_humble_wand Not enough to sayNothing 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 v2.1.271: on
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.
Read once, for one account on one subscription tier, against v2.1.271. It isn't a statement about your account. What a flag value here can and cannot tell you
New in this build: tengu_humble_wand