Unclear Whether the comparison changes what Claude Code does or is only recorded, and how "count" differs from "on", is not shown.
What
In remote sessions, Claude Code can stage a file, which means downloading a copy from a remote file store so it is available locally. New code can first ask the store for the file's details and compare them with the local copy:
- Store lookup: it calls
/v1/filestore/fs/readFileMetadatato get the file's size and md5 (a fingerprint of its contents) and compares them with the local file. - Local hashing: it computes the local file's fingerprint only when the local file is read-only (mode
0444), has a single hard link and has the same size as the stored file. - Verdict: the result is one of
same_contents,md5_differs,size_differs,local_file_changedand others. - Skipping: in mode
on, a matching file is not downloaded again and the step is recorded asalready_present. Before, the file was always fetched. - Pausing: if the check fails, it pauses for a set time or for the rest of the process (
stage_file_store_check_paused). - Modes: the mode comes from the
tengu_quiet_ledgergate, which accepts onlycount,onoroffand falls back tooffin the code. The staging step passes the gate value only whene.force === !0, and passesoffotherwise. Whatcountdoes differently fromonwas not traced. - Real path check: on Linux, the existing already-present check now confirms the file's real location through
/proc/self/fd.
Nothing has been read about the tengu_quiet_ledger or tengu_stage_file_completed gates, so whether this is active is not known.
Why
When the active mode is on, files that are already present and unchanged would not be downloaded again in remote sessions. The extra Linux path check makes the already-present decision rely on where the file really is.
tengu_quiet_ledger 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 v2.1.314: not a boolean we can read
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 v2.1.314. 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.
Whether the comparison changes what Claude Code does or is only recorded, and how "count" differs from "on", is not shown.
New in this build: tengu_quiet_ledger