Repository navigation
[FEATURE] Bring back the previous thinking was displayed (prior to version 2.0.0) #8371
Closed as not planned
Description
Preflight Checklist
- I have searched existing requests and this feature hasn't been requested yet
- This is a single feature request (not multiple features)
Problem Statement
The previous way in which thinking was displayed was great, I don't understand why you had to change that.
Also the TODOs require ctrl+, this brings less transparency to the user!
If any other companies should show the model thinking traces like you do.
Stop changing usability defaults without a way to rollback.
Proposed Solution
make the default behaviour toggleable.
Alternative Solutions
No response
Priority
Critical - Blocking my work
Feature Category
CLI commands and flags
Use Case Example
No response
Additional Context
No response
Reacted by alextranscend, heguro, Chris, normalnormie, Kioja Kudumu, Emil Yordanov, Trevor Bekolay, Marcelo Romero, janbam, Emanuele Sabetta and 59 more
Activity
kkudumu commentedon Sep 30, 2025
Came here looking for this as well...
Oichkatzelesfrettschen commentedon Sep 30, 2025
Ah here we go. I submitted feedback about thinking via /feedback and it pointed me here to report my issues. The thinking system is tab on and tab off with no immediately clear mode toggle as with using think, think hard, think harder prior for various levels. tab tab tab tab tab just toggles off and on instead of cycling through the levels, as expected.
I have one tab which has the prior version; and think, think hard, think harder continue to work there, so the issue is front-end UX rather than than a backend limitation.
tjx666 commentedon Sep 30, 2025
Yes, I want to know how claude thinking
Emasoft commentedon Sep 30, 2025
Yes! I really can’t work without being able to follow the model train of thought! 🙏
ctrlbrk42 commentedon Oct 2, 2025
Add a env variable to make it permanently visible
davidjson commentedon Oct 2, 2025
You can downgrade via
npm i -g @anthropic-ai/claude-code@1.0.128to display thinking if you don't rely on the new features.normalnormie commentedon Oct 4, 2025
"Stop changing usability defaults without a way to rollback."
Couldn't have said it better
tqwhite commentedon Oct 4, 2025
Showing text then hiding it is an awful experience. I see stuff, start reading it, then it goes away, I scroll up to see if it's still there, the display changes below so I am confused.
Moving text around the screen is horrible. I'm reading something then it zips up the terminal window. Where was I? I have to reread the thing to find my place. I lost time and am confused.
With the previous version, while Claude was working, I would read the thinking and activities, note inconsistencies or get new ideas. I could interrupt and offer guidance. It made Claude a beautiful, interactive tool. I not only had control but felt confident I know what was happening.
Now it's a black box. You killed the pleasure.
Revert these changes. The tool is ruined.
atrawog commentedon Oct 4, 2025
This new gui change is
press Ctrl+O to show thinkingstupid. Why dear Anthropics has nobody learn from the task list not showing up complains?People WANT and NEED inside into what Claude is doing not
press Ctrl+O to show thinkingGUI chances the serve nopress Ctrl+O to show thinkingpurpose at all.OmniZenRaj commentedon Oct 5, 2025
New CLI needs to stop auto hiding parts of what I am trying to read. Please revert to pre 2.0 version of CLI output. Thanks
Nordwolf commentedon Oct 5, 2025
I think it's bad when mass-market companies do opinionated changes that do not fit a large part of their userbase - but that is at least "justifiable" when the users are the product.
However when it's a professional. paid tool - such opinionated changes ruin workflows and productivity. If you want to introduce a feature to hide thinking - make it a feature, even make it default, but do not take away the option that is so widely used.
aleks-apostle commentedon Oct 7, 2025
I wrote a simple script that toggles the expanded thinking traces by default until Anthropic patches this:
https://github.com/aleks-apostle/claude-code-thinking-patch
thomas-schweich commentedon Oct 17, 2025
#8477 and #9099 do appear to be genuine duplicates. Which one should we be looking at/reacting to upvote this feature request? I'm commenting here because it's the lowest issue number.
Chain-of-thought/thinking is one of the things that makes Claude models stand out (though the newer GPTs are pretty good at it too now). I'd think Anthropic would want to show it off. I like that I can quickly tell if the model is going off the rails by monitoring it. I was definitely disappointed to see it go in the 2.0 version of Claude Code.
tjx666 commentedon Oct 19, 2025
I highly doubt the Claude code team can't see the GitHub issues. They probably just work on features they are interested in, and after finishing them, they search if there are similar requests and then close the issues.
gruntlord5 commentedon Oct 27, 2025
I would also like this added, its pretty critical to seeing how the AI is going to execute what I ask vs what it tells me
github-actions commentedon Dec 8, 2025
This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.
saeedesmaili commentedon Dec 14, 2025
+1, don't close this until it's properly resolved by anthropics.
github-actions commentedon Jan 15, 2026
This issue has been automatically closed due to 60 days of inactivity. If you're still experiencing this issue, please open a new issue with updated information.
marcindulak commentedon Jan 15, 2026
This issue was closed incorrectly despite recent human comments. This behavior of the bot is reported at #16497. Please upvote that issue, so maybe it gets noticed.
anthrotype commentedon Jan 15, 2026
yes, please keep this open (as well as the dozens of other similar/duplicate issues about bringing back the old inline thinking)!
in the meatime, I asked Claude to write a skill to keep this patch updated every time a new claude code is released: aleks-apostle/claude-code-patches#9
saeedesmaili commentedon Jan 15, 2026
I guess Anthropic doesn't care, their policy is to close any issue after 60 days!
marcindulak commentedon Jan 15, 2026
I think Anthropic is unaware of the #16497 problem.
The current behavior of the bot is to close any issue that has "This issue has been inactive for 30 days" bot warning, independently whether a human comments are present afterwards.
There is still a hope that this behavior of the bot will be changed if anyone from Anthropic reads the issue.
github-actions commentedon Jan 23, 2026
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.