Claude Code can now split the text of Claude's replies into pieces, each linked to the web search result it came from. It only does this when both of these hold:
- the environment variable
CLAUDE_CODE_WEBSEARCH_CITATIONSis set - a remote switch is on; it is off when the server sends nothing
Each link is checked against the search results that actually came back in the conversation, and the number of links per message is capped. That switch has read off both for this site's account and for an anonymous check, in readings not yet taken under this release.
This prepares replies that point to their web sources piece by piece, and setting the environment variable on its own does nothing while the switch is off.
tengu_jolly_sparrow_enabled Off in both readingsThe flag server returned off for the account this site reads and for the anonymous baseline. A reading of off cannot rule out a rollout these two readings sit outside of.
This account: off · anonymous baseline: off · compiled default in v2.1.309: 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.
Read once, for one account on one subscription tier, against v2.1.309. It isn't a statement about your account. What a flag value here can and cannot tell you
A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.
New in this build: tengu_jolly_sparrow_enabled