If you use Claude Code daily, you know the problem: you explain your project, your conventions, your preferences to Claude – and in the next session, everything starts from scratch. Until now, you had to maintain the /memory file manually.
As of February 26th, that’s history.
Auto-Memory: Claude Learns Along
Claude Code now automatically saves useful context to the memory file. Project structure, recurring patterns, your preferences – everything Claude considers useful ends up in long-term memory. You can review and edit this at any time via /memory.
Sounds simple. But it’s a game changer for anyone who uses Claude Code as a daily tool. Instead of starting each session at zero, Claude builds on what it has already learned about your project.
/copy: Selectively Copy Code
The second update solves a small but annoying everyday situation: Claude produces a response with multiple code blocks – and you only want to copy a specific one. The new /copy command shows you an interactive picker where you can select individual blocks or the entire response.
Smarter Bash Prefixes
Under the hood, the “Always Allow” logic for bash commands has also been improved. For compound commands like cd project && npm install && npm test, Claude now calculates smarter prefixes per subcommand – instead of treating the entire command as one unit. Fewer unnecessary prompts, smoother workflow.
Bottom Line
No revolution, but exactly the kind of updates that noticeably improve the daily workflow. If you use Claude Code, you should update to the latest version – the memory feature alone is worth it.
Source: Claude Code Changelog