Claude Code v2.1.111 adds Opus 4.7 effort levels, auto mode for Max subscribers, and cloud-based code review

Anthropic released Claude Code v2.1.111 on April 16, introducing Claude Opus 4.7 with three configurable effort levels. The new 'xhigh' level sits between the existing 'high' and 'max' settings, allowing developers to tune the balance between response speed and model intelligence via the /effort command. An interactive slider with arrow-key navigation was added when /effort is called without arguments.

Auto mode became available for Max tier subscribers using Opus 4.7. This mode automatically adjusts behavior based on context and no longer requires the --enable-auto-mode flag to activate. A new /theme command adds an 'Auto (match terminal)' option that synchronizes the tool's appearance with the user's terminal dark or light mode preference.

The /ultrareview command enables running comprehensive code reviews in the cloud using parallel multi-agent analysis. Developers can invoke it without arguments to review the current branch or pass a GitHub PR number to fetch and review a specific pull request.

Several operational improvements target common workflows. Read-only Bash commands with glob patterns and cd commands no longer trigger permission prompts. A new /less-permission-prompts skill scans transcripts for read-only tool calls and proposes a prioritized allowlist for configuration. Windows users can progressively access the PowerShell tool via the CLAUDE_CODE_USE_POWERSHELL_TOOL environment variable, with Linux and macOS support available when pwsh is in the PATH.

Bug fixes address terminal stability, LSP diagnostics ordering, and session management. Terminal display tearing in iTerm2 with tmux was resolved. The tool no longer re-scans entire projects for file suggestions in non-git directories and correctly preserves session names after the /clear command.