You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Add Config Option to Disable Automatic Context Compaction in Codex CLI #11716
Please add a user-configurable way to fully disable automatic context compaction in Codex CLI (for example, model_auto_compact_enabled = false in config.toml). The current behavior appears to always auto-compact once usage crosses a threshold, which can interrupt workflows that depend on preserving full, verbatim conversation history for debugging, audits, or deterministic reproduction. I’m not asking to change the default behavior—keeping auto-compaction enabled by default is reasonable—but advanced users should have an explicit opt-out switch and clear documentation of the tradeoffs.
Activity
etraut-openai commentedon Feb 13, 2026
We're already tracking this feature request in #4106. Please upvote that issue.