anthropics/claude-code v2.1.210

July 16, 2026 feature_update 313 words

Claude Code v2.1.210 Brings UX Improvements and Critical Security Fixes

TL;DR

Live feedback: Tool execution now displays real-time elapsed-time counters, eliminating the appearance of frozen operations
Security hardened: Fixed critical isolation bypass where subagents could mutate the main repository, plus patched permission rule vulnerabilities
Polish push: Multiple UI and integration fixes address external editor compatibility and session stability issues

What happened

Anthropic released v2.1.210 of Claude Code, its AI-powered development toolkit, rolling out a maintenance update focused on user experience refinement and security hardening. Available on GitHub, the release addresses nine distinct issues spanning interface responsiveness, permission enforcement, and system reliability.

The most visible change targets a longstanding UX pain point: long-running tool executions now display animated elapsed-time counters on the collapsed tool summary line. This addresses user feedback that extended operations appeared unresponsive, improving confidence in tool execution status.

On the security front, the update patches a meaningful isolation vulnerability where subagents configured with isolation: 'worktree' could execute git-mutating commands against the main repository checkout instead of their isolated workspaces. The release also adds startup warnings for deprecated permission rules (Write(path), NotebookEdit(path), Glob(path)), nudging developers toward safer alternatives like Edit(path) and Read(path).

Additional fixes target the ultracode keyword opt-in mechanism, which was incorrectly firing on non-human inputs like webhook payloads and relay PR comments, potentially triggering unintended automation. The update also resolves text rendering and clipboard issues that caused È/É characters to leak into external editors and UI components to contaminate crash telemetry.

Session reliability received attention with improvements to the claude attach command, which previously failed intermittently during daemon transitions with "job not found" or "agent is still starting" errors.

What happens next

Users should upgrade to capture the security isolation fix immediately. The quality-of-life improvements—particularly the live elapsed-time feedback—enhance developer workflows, while permission deprecation warnings signal breaking changes likely coming in future releases. This article does not contain affiliate links.