Auto-compact is not working like previously
Open 💬 2 comments Opened Jan 23, 2026 by King-718
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
0.89.0
What subscription do you have?
api
Which model were you using?
gpt-5.2 high/xhigh
What platform is your computer?
macos
What terminal emulator and version are you using (if applicable)?
ghostty
What issue are you seeing?
Anyone else notice regression in Auto-compact ? version 0.86 and before auto-compact was working beautifully and agent would pick up from where it left off. Version 0.87 onwards the agent has started to get confused and lost. also seen on reddit https://www.reddit.com/r/codex/comments/1qk55js/why_does_it_keep_forgetting_what_it_did_before/
What steps can reproduce the bug?
Revert auto-compact
What is the expected behavior?
Auto-compact was one of the best parts previously, no other agentic system matched it compared to codex.
Additional information
_No response_
2 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Yes, I mean, when I used /compact, the model just forgot everything. The context become 100% again.