[Release] anthropics/claude-code v2.1.205: v2.1.205

Claude Code v2.1.205 introduces critical bug fixes and improvements across session management, file handling, and auto mode functionality. Key enhancements include blocking tampering with session transcripts, fixing JSON schema validation issues, improving background agent state tracking, and optimizing memory usage during auto-updates. The release also strengthens security by preventing fabricated approvals and adds better diagnostics with the enhanced `/doctor` command.

Key Points

• •Added auto mode rule to block tampering with session transcript files for enhanced security
• •Fixed JSON schema validation: invalid schemas no longer silently produce unstructured output, and `format` keyword is now properly supported
• •Fixed message loss when Claude reaches `--max-turns` limit by ensuring messages sent during work are preserved
• •Fixed Windows worktree removal bug that could delete files outside the worktree when NTFS junctions or symlinks existed
• •Improved background agent state tracking: agents no longer incorrectly show as 'failed' or 'completed' after resumption with `SendMessage`
• •Fixed `claude attach` command to wait for background agents mid-upgrade instead of erroring
• •Enhanced auto mode security: now asks before running `rm -rf` on unresolved variables
• •Optimized auto-update memory usage by streaming binary downloads to disk instead of buffering (~400 MB reduction)
• •Improved agent view with colored state indicators, classifier-written headlines, and better PR linking for existing pull requests
• •Upgraded `/doctor` command to full setup checkup with diagnostic and fix capabilities; `/checkup` is now its alias

Workflow Diagram

Concepts