Follow Discord
Sweep 25 Sep 2026 · 19:33Z Build v2.1.283 504 read Stable v2.1.274 Latest v2.1.283 Next v2.1.283 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.281 ·

Directory sync tolerates malformed git history and caps the set-aside list

Directory sync now accepts repos with known malformed commit data, dates fork points oldest first and caps its setAside list

Group of 2 Under the hood Internal Changes
JSON All of v2.1.281
Under the hoodTier: how much it should matter to you
1Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
Cloud SyncArea: what it touches
Internal ChangesKind: in v2.1.281,
Internal ChangesSection of the release

What

When a cloud session and your machine exchange git history, Claude Code checks each incoming bundle with index-pack. It used to run this check in plain --strict mode. That changes in several ways:

  • index-pack now runs with --strict= plus a list of known problems to ignore, and the list depends on which side is receiving. Both the container and your machine ignore problems such as badDate, badEmail and zeroPaddedFilemode, and the container also ignores gitmodulesUpdate.
  • A bundle that is still refused is logged as dir_sync_git_receive_fsck_refused.
  • Creating a bundle adds --no-reuse-object, depending on your git version.
  • When there are more fork points than the limit allows, they are chosen oldest first, using dates from rev-list --timestamp. A fork point is where two lines of history split.
  • Git now runs from a program location found for the session's root folder, not from one fixed program.
  • The sync data gains a setAside list and a setAsideTruncated marker. A new helper caps setAside at a limit and sets setAsideTruncated when the list is cut short.

Why

Repositories whose history contains malformed commit details, such as bad dates or email addresses, should no longer fail to sync. The set-aside list also can no longer grow without bound.

How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtThe finding does not say what the set-aside list holds, where a reader would see it, or what the limit is.

See this entry in the whole of v2.1.281 →

Feedback