Add support for /compact
Resolved 💬 7 comments Opened Oct 8, 2025 by rucinski-ttp Closed Jan 21, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of the VS Code extension are you using?
0.4.19
Which IDE are you using?
VS Code
What platform is your computer?
Microsoft Windows NT 10.0.26200.0 x64
What steps can reproduce the bug?
After using the tool for a short while, the conversation is unable to continue, with the error Codex ran out of room in the model's context window. Start a new conversation or clear earlier history before retrying.
What is the expected behavior?
Similar to the CLI tool, the chat should be condensed and summarised to free up space and allow the conversation to continue. This would be either automatic when the context window limit is reached, or manually through a button.
What do you see instead?
When this limit is hit, the conversation has to be abandoned and a new chat started.
Additional information
_No response_
7 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Guys I really need this, Please if possible pick this on priority. This is the only feature that is keeping me away from using codex.
I don't think they've added /compact as manual command, butthey have added automatic compaction into the vs code extension, does that help?
I use it in cursor, but I'm good,
gpt-5.1-codex-maxdoes compaction automatically. Thank youI see there's a /compact command now, close?
we don't have /compact command yet, models themselves do the compaction now https://openai.com/index/gpt-5-1-codex-max/
@matixlol
With the latest models, compaction is now performed automatically. Manual compaction should no longer be needed, so we're unlikely to expose a manual compaction command in the IDE extension. The command is still available in the CLI, but we will likely deprecate that eventually. Our long-term goal is to eliminate the need for developers to think about compaction and context window management.