Source Intelligence

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

All of v1.0.58 Home All releases olderv1.0.57 v1.0.59newer
Claude Code v1.0.58

New Features

Agent Color Coding

  • Agents now have color-coded identification in the interface
  • New color palette: claude, success, bashBorder, planMode, autoAccept, warning, permission, error
  • Color assignments persist per agent type for consistency
  • The general-purpose agent remains uncolored

PDF File Support

  • Added native PDF reading capability to the Read tool
  • PDF files up to 32MB can now be processed directly
  • PDF content is displayed visually as Claude Code is multimodal

Enhanced Permission Messages

  • New DETAILED_PERMISSION_MESSAGES environment variable
  • When enabled, provides more context about which specific commands are pre-approved
  • Permission prompts now show reminders of allowed commands for each tool type

OAuth 2.0 Client Authentication

  • Added comprehensive OAuth 2.0 client authentication support
  • Supports multiple authentication methods:
  • client_secret_basic (HTTP Basic Auth)
  • client_secret_post (form-encoded)
  • none (public clients)
  • New error handling for OAuth flows with specific error types

Bash Command Prefix Support

  • New CLAUDE_CODE_BASH_PREFIX environment variable
  • Allows prefixing all bash commands with custom wrapper scripts
  • Useful for sandboxing or monitoring command execution

See this entry in the whole of v1.0.58 →