Claude Code v2.1.193 Changelog

This update is all about making Claude Code—the tool that lets the Claude AI work directly in your computer's terminal—smarter, safer, and much better at multitasking. It is especially important for developers who want Claude to run tasks in the background without getting confused, and for anyone who wants clearer control over what Claude is allowed to do on their computer.

What's new

• Smarter safety checks for commands: You can now set Claude to check *every* terminal command you run to make sure it is safe, rather than just checking commands that look risky.
• Clearer block reasons: If Claude blocks a command from running because of your security settings, it will now tell you exactly *why* it blocked it in your chat history, notifications, and permissions menu.
• File path autocomplete: When you type terminal commands using the ! shortcut, Claude will now suggest file paths as you type, just like a regular terminal does when you press the Tab key.
• Login reminders for extra tools: If Claude connects to external tools (called MCP servers) that require you to log in, it will now warn you right when it starts up and tell you where to go to log in.
• Automatic memory saving: If you have terminal commands running in the background that you aren't using, Claude will now automatically shut them down if your computer starts running out of memory (RAM).
• Response logging for developers: For teams tracking how the app performs (using "telemetry" or activity logging), Claude can now log its own text responses. If you already log what users type, this will start logging Claude's replies too, though you can easily turn this off to keep replies private.

Improvements

• Better multitasking: When you start a background task, Claude no longer thinks it has to stop talking to you. It now knows it can keep chatting and helping you with other things while that background task runs.
• Auto-reconnecting to tools: If Claude loses its connection to an external tool because your login expired, it will now automatically try to log back in and reconnect for you.
• Automatic plugin renaming: If a plugin you use gets renamed in the extension store, Claude will automatically update your settings to the new name so nothing breaks.
• Friendlier folder messages: If you try to add a folder that Claude is already working in, it will now give you a much clearer, friendlier message.

Fixes

• No more blank screens after login: Fixed a bug where logging in would temporarily show blank or outdated information in menus like /model instead of updating right away.
• Fixed false alarms when backgrounding: Fixed an issue where sending tasks to the background would trigger a false warning saying your tasks were about to be lost, even though they were perfectly safe.
• Quiet updates for background helpers: Background "agents" (mini-Claudes working on tasks) will no longer get confused after an automatic software update and repeatedly ask you if they should resume.
• Banished "ghost" agents: Fixed a bug where putting your main conversation into the background created a duplicate "ghost" assistant that would accidentally repeat the whole conversation from the beginning.
• Better sidebar visibility: In the panel where you see your active AI helpers, looking at a sub-helper will no longer accidentally hide all the other helpers you have running.

Breaking changes

No breaking changes in this release.