Claude Preferences

Open Preferences -> AI Integration -> Claude. Enable Claude integration (will be available in Graph popup menus) makes Claude run actions available.

• Path selects the installed Claude executable or a custom launcher. You can also enter a command name resolved from PATH.
• New Arguments supplies the arguments for starting a conversation.
• Resume Arguments supplies the arguments for continuing an existing session. A native invocation uses ${sessionId} to identify that session.
• Run Claude in a dedicated SmartGit worktree controls worktree isolation.
• Model and Effort Level, under Defaults, set the initial model settings for conversations. Choose from the suggestions or enter values supported by your installed agent and model.
• Sandboxing selects the permissions described under Sandboxing.
• Cleanup Arguments, under Custom containerized Claude, optionally invokes the launcher to clean up a deleted session’s environment.

SmartGit supplies prompts on standard input. Each argument field has a Variables menu for inserting launcher placeholders.

Auto-Configure replaces the path and argument templates and restores the model defaults. To apply the setup, select the acknowledgement checkbox and click Allow and Apply. On native Windows, the acknowledgement states that Claude will run unrestricted and without a firewall.