[Release] anthropics/claude-code v2.1.232: v2.1.232

Claude Code v2.1.232 introduces major improvements to subagent forking, cross-session messaging, and security hardening. Key features include default subagent forking with prompt cache inheritance, @-mention syntax for direct session communication, GitLab marketplace support, and numerous security fixes for permission bypasses and credential handling. The release also enhances Remote Control stability, improves managed settings validation, and fixes various UI and performance issues.

Key Points

• •Subagent forking enabled by default: subagents inherit full conversation and prompt cache; non-teammate agents run in background by default
• •@-mention syntax allows direct messaging between Claude sessions using SendMessage without confirmation
• •Interactive sessions maintain unique names automatically with `name-word-word` variants to prevent conflicts
• •GitLab support added to plugin marketplaces with nested subgroup support and auth-failure hints
• •Enhanced secret redaction for GitLab tokens (glrt-, gloas-, glptt-, glagent-, glimt-, glsoat-, glcbt-, glft-, glffct-) and full redaction of routable tokens
• •Security fixes: PowerShell parameter bypass, Windows Git Bash symlink bypass, nested git repository trust inheritance, and mTLS certificate rotation
• •Remote Control improvements: session reattachment, conversation history restoration, 30-minute reconnection window after network blips
• •Gateway validation hardened: empty marketplace groups, malformed email domains, and unknown Desktop settings now fail at boot
• •Improved fullscreen streaming performance by eliminating full conversation re-normalization on every update
• •Fixed MCP connection hangs, cross-session credential inheritance, and filesystem sandbox protected-path bypass

Workflow Diagram

Concepts