Claude Code 2.1.234: Mid-Turn Controls and Security Fixes
This episode breaks down Claude Code 2.1.234’s biggest upgrades, including mid-turn directory changes, on-demand documentation loading, and auto-resume for daily limits. It also covers a critical Windows path security fix and the guardrails that keep TUI mode changes locked down when tool restrictions are in place.
Chapter 1
Mid Turn TUI Dialogs in Claude Code Version 2.1.234
James Turner
So, version 2.1.234 of Claude Code just landed, and, uh, the big shift here is that you no longer have to kill a running turn just to change your environment.
Lachlan Reed
Mate, you have no idea how much time that saves me in the shed. I, I, I was halfway through a multi repo refactor yesterday, right? Claude was hammering away on the backend, and suddenly it needed context from dot slash packages slash shared. Old days, you had to cancel the whole execution, re run it, burn those tokens all over again.
James Turner
But now you can just invoke slash add dir mid execution?
Lachlan Reed
Exactly! You hit slash add dir, point it to dot slash packages slash shared, or pop open slash permissions or slash config right while the tool is running. It just dynamically attaches the new directory to the scope and keeps rolling. No restart needed, flat out brilliant.
James Turner
That, that, that whole idea of not wasting context connects directly to what they did with the built in claude api skill, too. Previously, if you enabled that skill, it would do an eager full spec prompt injection. That was dumping over 200,000 tokens into your context window right off the bat!
Lachlan Reed
200,000 tokens before you even type a prompt? Crikey, that is a proper memory hog.
James Turner
Right! But in 2.1.234, they switched to on demand documentation loading. So that initial context overhead drops from over 200,000 tokens down to roughly 25,000 tokens. That is nearly an eightfold reduction in idle context cost.
Lachlan Reed
That is massive for token budgets, especially when you are running longer sessions. And, uh, speaking of long sessions, they added auto resume for daily usage limits, yeah?
James Turner
Yeah, exactly. If you hit your daily limit on claude.ai, the session doesn't just crash out anymore. It pauses, waits for the reset window, and then automatically continues where it left off. You can toggle that on or off in slash config.
Lachlan Reed
That is so handy if you leave a huge batch job running overnight. But, wait, there was also a security patch tucked into this release, wasn't there? Something about Windows path handling?
James Turner
Yeah, critical fix there. It rejects Windows NT namespace paths, the ones starting with backslash question mark question mark backslash. Turns out attackers could use those paths in remote file reads or included CLAUDE.md files to leak NTLM hash credentials over SMB.
Lachlan Reed
Oof, leaking NTLM credentials just from loading a repository file, that is a nasty vector. Good on them for patching that out fast.
James Turner
Definitely. Now, there is one caveat to keep in mind if you are messing with slash TUI mode switches. If you launched your session with explicit tool restriction flags, like allowed tools or disallowed tools, slash TUI will outright refuse to switch modes.
Lachlan Reed
Ah, so it keeps you locked into your security sandbox so you don't accidentally bypass your own rules mid stream.
James Turner
Precisely. Safety guardrails stay intact, but you get way more control mid turn. Good update overall.
Lachlan Reed
Too right. Well, I am off to run some more mid turn add dirs. Catch you later, James.