Version 1.0.37

# Claude Code v1.0.37 Changelog

• Added support for -p / --print command-line flags and automatic detection of non-TTY output
• When running claude -p or claude --print, or when output is piped/redirected, Claude will operate in print mode
• This allows for better integration with scripts and command pipelines

Automatic Config Backup and Recovery

• Configuration files are now automatically backed up when corruption is detected
• If the main config file is corrupted, a timestamped backup is created (e.g., config.json.corrupted.1234567890)
• Users are notified of backup locations and provided with recovery instructions
• Example recovery message:

  Claude configuration file at ~/.claude/config.json is corrupted
  The corrupted file has been backed up to: ~/.claude/config.json.corrupted.1234567890
  A backup file exists at: ~/.claude/config.json.backup
  You can manually restore it by running: cp "~/.claude/config.json.backup" "~/.claude/config.json"

Improved Config File Locking

• Added file locking mechanism for configuration updates to prevent concurrent write conflicts
• If locking fails, the system gracefully falls back to non-locked saves with appropriate logging

• New environment variable CLAUDE_CODE_DISABLE_COST_REPORTING to disable cost reporting features
• Usage: export CLAUDE_CODE_DISABLE_COST_REPORTING=true

• Added proper error messaging when an agent's proposed plan is rejected by the user
• Provides clearer feedback in the interactive workflow

• Added new AbortError class for better error categorization and handling of aborted operations

• Migrated from generic stream imports to more specific PassThrough stream for improved performance

• Added new permission request structure for tools, including:
• Tool name identification
• Input validation
• Unique tool use request IDs for tracking

• Fixed configuration reading and writing logic to be more robust
• Improved error messages and user guidance for configuration issues
• Enhanced logging throughout the configuration management system

• Refactored configuration file operations with better error handling and recovery
• Added comprehensive logging for debugging configuration issues
• Improved modularization of stream processing components