At a glance: Claude Code v2.1.147 improves background sessions, introduces advanced code review features, and fixes over 30 bugs in enterprise security, shell integration, and platform-specific issues.
The new version of Claude Code brings significant improvements for agent development work. The update focuses on stabilizing background sessions, advanced code review capabilities, and fixing numerous bugs across various system areas.
Claude Code v2.1.147 introduces a series of optimizations that substantially improve the developer experience. Pinned background sessions now remain active in idle state, restart upon Claude Code updates, and are only terminated under memory pressure after unpinned sessions.
The previous /simplify function has been renamed to /code-review and now offers detailed reports on correctness issues at various effort levels. With the –comment flag, insights can be integrated directly as inline comments in GitHub pull requests.
The auto-update system has been fundamentally overhauled: it now automatically retries transient network errors, reports specific error categories and operating system error codes when issues occur, and displays the current version upon failed updates.
Further important improvements include optimized rendering performance for large file changes and more intelligent prompt history management that no longer stores consecutive duplicates.
The update also addresses critical security issues in enterprise login restrictions and fixes numerous bugs in areas such as shell snapshots, PowerShell integration, and plugin agent management. Windows users benefit from improved script validation and corrected behavior when handling NTFS junctions.
Source: github.com