Claude Code · Source-level changelog

Version 1.0.68

Package @anthropic-ai/claude-code Diff v1.0.67 → v1.0.68
11
Features & Changes
0
Bug Fixes
0
In Development
0
Env Vars / Flags

Changes

8 items

Based on my analysis of the diff file for version 1.0.68, here is the detailed changelog:

# Claude Code v1.0.68 Changelog

Teleport - Resume Sessions Across Machines #

A powerful new --teleport feature allows you to resume Claude Code sessions from other machines or continue someone else's shared session.

Usage:

# Resume by session ID
claude --teleport abc123def456

# Resume from a URL
claude --teleport https://claude.ai/chat/abc123def456

Features:

  • Automatically fetches conversation history from remote sessions
  • Switches to the appropriate git branch if needed
  • Offers to stash uncommitted changes before switching branches
  • Supports custom headers via TELEPORT_HEADERS environment variable
  • Validates authentication status before proceeding

Environment Variables:

  • TELEPORT_RESUME_URL: Override the default URL for session resumption
  • TELEPORT_HEADERS: Pass custom headers as JSON (e.g., export TELEPORT_HEADERS='{"Authorization": "Bearer token"}')

Customizable Output Modes #

Configure how Claude Code communicates with you by selecting from different output modes.

Features:

  • Support for custom output modes stored in .claude/output-modes/ directories
  • Both user-level and project-level output mode configurations
  • Interactive selection UI with up to 10 visible options (increased from 3)
  • Dynamic loading with "Loading output modes..." indicator
  • Each mode includes a description to help you choose

Enhanced Hooks Management #

New global control over Claude Code's hook system for better troubleshooting and control.

New Capability:

  • Disable all hooks globally without deleting your configuration
  • Clear UI indicator when hooks are disabled showing:
  • Warning banner explaining hooks are disabled
  • Count of configured hooks that aren't running
  • Option to re-enable all hooks with one click

Expanded Shell Command Support #

10 new commands added to the safe command whitelist, allowing execution without approval prompts:

  • nl - Number lines in files
  • hostname - Display system hostname
  • arch - Show system architecture
  • groups - List user groups
  • nproc - Display number of processors
  • basename - Extract filename from path
  • dirname - Extract directory from path
  • realpath - Resolve to absolute path
  • base64 - Encode/decode base64 data
  • file - Determine file type

️ Better Error Handling #

  • Improved shell execution with shell: true option for better compatibility
  • Enhanced error messages with clearer explanations
  • Try-catch blocks added to critical command execution paths
  • Better timeout and cancellation handling for long-running commands

UI/UX Enhancements #

  • Loading states with descriptive messages throughout the interface
  • Interactive prompts for git stash decisions during teleport
  • Clearer permission descriptions for tool access levels
  • Improved session selection UI when multiple sessions are available

Removed Features #

  • Removed NotebookRead tool and related Jupyter notebook reading functionality
  • Cleaned up unused imports for node:stream

Command Execution Improvements #

  • New command tracking infrastructure with unique command IDs
  • Enhanced verbose logging with timestamped output
  • Escaped command strings for improved security
  • Better handling of piped commands and multi-line output

Breaking Changes

None - This release maintains backward compatibility with existing workflows.

Notes

3 items
  • The teleport feature requires a clean git working directory or will prompt to stash changes
  • Custom output modes can be placed in .claude/output-modes/ in your home or project directory
  • The expanded shell command whitelist improves productivity for common system tasks while maintaining security