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 v0.2.80 Home All releases olderv0.2.79 v0.2.81newer
Claude Code v0.2.80

New Features

OAuth Support for Claude.ai Pro/Team Users (CLAUDE_MAX)

  • New environment variable: CLAUDE_MAX enables OAuth authentication for Claude.ai Pro/Team users
  • When enabled, adds user:inference scope to OAuth permissions, allowing Claude Code to use your Claude.ai account for API access
  • OAuth flow automatically handles token refresh when tokens are near expiration (within 5 minutes)
  • Separate token storage for Claude.ai OAuth vs Console API keys

Deny Rules for Tool Permissions

  • New permission behavior: In addition to "allow" rules, you can now create "deny" rules
  • Navigate between Allow/Deny modes using Tab or arrow keys in the permission rules interface
  • Denied tools will be automatically rejected without prompting
  • Example: Add a deny rule for Bash to prevent any shell command execution

Managed Settings Support

  • Enterprise policy control: System administrators can now configure tool permissions via managed settings
  • Managed settings are loaded from system-level policy files:
  • macOS: /Library/Application Support/ClaudeCode/policies.json
  • Linux: /etc/claude-code/policies.json
  • Managed permission rules cannot be modified or deleted by users
  • Rules are displayed with a special indicator showing they're managed by your organization

Tool Renaming

  • dispatch_agent renamed to AgentTool for better clarity
  • ReadNotebook renamed to NotebookReadTool for consistency
  • LS renamed to LSTool for consistency

See this entry in the whole of v0.2.80 →