Agent DailyAgent Daily
releaseintermediate

[Release] anthropics/claude-code v2.1.132: v2.1.132

By ashwin-antgithub
View original on github

Claude Code v2.1.132 is a maintenance release focused on stability, terminal handling, and user experience improvements. Key fixes address graceful shutdown behavior, emoji/character rendering issues, clipboard operations, IDE terminal compatibility, and MCP server reliability. The release also adds environment variables for session tracking and alternate-screen rendering control, plus improvements to slash command UI and error messaging.

Key Points

  • Added `CLAUDE_CODE_SESSION_ID` environment variable to Bash tool subprocess for session tracking consistency
  • Added `CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN=1` option to disable fullscreen renderer and preserve terminal scrollback
  • Fixed external SIGINT handling (IDE stop buttons, kill -INT) to properly restore terminal modes and show resume hints
  • Fixed emoji/character rendering issues: emoji truncation corruption, mid-grapheme cursor positioning, decomposed accent character handling in vim
  • Fixed clipboard operations: pasting text starting with `/`, escape sequence interleaving with bracketed paste, and `/usage` screenshot copying on Linux/X11
  • Improved IDE terminal compatibility: fixed mouse wheel scrolling in Cursor/VS Code, JetBrains 2025.2 arrow key spurious events, and Windows Terminal Shift+Enter messaging
  • Fixed MCP server reliability: unbounded memory growth from non-protocol stdout, silent tool fetch failures now retry with status feedback
  • Fixed session resumption issues: `--permission-mode` flag being ignored, plan mode not re-applying after ExitPlanMode, dead keyboard input on Windows
  • Enhanced UI consistency: slash command autocomplete scaling with terminal height, statusline context_window token accuracy, `/effort` picker respecting env var overrides
  • Fixed fullscreen mode blank screen after sleep/wake/Ctrl+Z and Alt+T thinking toggle on macOS terminals without Option-as-Meta

Found this useful? Add it to a playbook for a step-by-step implementation guide.

Quality

Concepts