When prime "backtrack" mode, context is often lost

Resolved 💬 3 comments Opened Nov 4, 2025 by BiFangKNT Closed Jan 17, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of Codex is running?

codex-cli 0.53.0

What subscription do you have?

Business

Which model were you using?

gpt-5-codex

What platform is your computer?

Linux 6.8.0-87-generic x86_64 x86_64

What issue are you seeing?

As the title says, after chatting for a while, when I invoke the “backtrack” mode and fork the conversation, the context is mysteriously lost—this happens frequently.

What steps can reproduce the bug?

Looking at docs/getting-started.md#escesc-to-edit-a-previous-message, I believe the issue lies in the “trim the visible transcript” step.

Context window: 100% left (4.43K used / 272K)

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 9 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #5333
  • #5957
  • #6085

Powered by Codex Action

BiFangKNT · 9 months ago

saw:https://github.com/openai/codex/issues/5333#issuecomment-3419729561
but didn't work

etraut-openai contributor · 7 months ago

Closing out old bugs.