Anthropic’s changelog lists Claude Code 2.1.329, dated 19 November 2026. It is a large maintenance release (the changelog counts 91 bug fixes), but a few changes affect how teams work.
Changes worth knowing
- Whole-branch review.
/code-reviewwith no target, on most models, now reviews the whole branch against the default branch. It also fixes a bug where multi-commit branches lost the diff. - Clearer failures.
/code-reviewgiven a pull request number or URL now stops, and says so, if it cannot read the pull request, instead of sometimes reviewing the checked-out branch. - Scheduled tasks. In new signed-in sessions, most scheduled tasks saved by earlier versions need scheduling again.
/tasksnow also lists a session’s scheduled tasks and forks. - Hooks and trust. A trust dialog now warns where a folder’s hooks are declared.
- Code Mode. Several fixes and limits, including a cap on a runaway loop that could fill memory.
- Stability. Fixes for sign-outs on macOS, memory use in long sessions and a settings file being reset after a network stall.
Why it matters
Anyone running automated Claude Code jobs should check that their scheduled tasks still exist after updating. Silent loss of a routine is the sort of failure that is only noticed weeks later. The review change is a reminder that AI tooling behaviour shifts between versions, so what “review” means this month may differ from last.
So what
If your supplier relies on AI tools, ask how they pin and test tool versions and who checks automated jobs still run. Our AI-assisted development work treats that as part of delivery. To discuss a build, start a project.