Codex CLI: Organize Threads into Sections & Browse Transcripts Faster
CodexCodex CLI 0.147.0 introduced persistent, manually ordered sections that let developers group related conversations together instead of scrolling through one long, flat thread list. The release also adds incremental loading for long transcripts, so large sessions open and scroll faster instead of loading the entire history at once. Together these changes extend the thread-management work Codex CLI shipped over the prior two releases — search and persisted names in 0.145.0, then pinning, forking, and side conversations in 0.146.0 — into a more structured, scalable way to manage a large volume of ongoing work.
Key Takeaways
- Persistent sections let developers manually group and order related conversations, replacing a single flat, chronological thread list.
- Incremental transcript loading means long, tool-heavy Codex sessions now load progressively while scrolling instead of all at once, cutting wait times.
- This is the third consecutive release to expand Codex CLI's thread-management system, following persisted names/search in 0.145.0 and pinning/forking/side-conversations in 0.146.0.
- Sections are user-defined, not automatic — developers choose the grouping and ordering themselves rather than relying on recency sorting.
- The feature targets developers running many concurrent Codex sessions, a workflow that has grown more common as Codex has added longer-running and multi-agent capabilities.
- Ships in Codex CLI 0.147.0, installable via
npm install -g @openai/codex@0.147.0, released August 7, 2026.
Sources & Mentions
2 external resources covering this update
Persistent, Ordered Sections for Conversations
Codex CLI 0.147.0, released August 7, 2026, gave developers a structured way to organize their growing list of conversations. Instead of one continuously scrolling thread list, Codex now supports persistent, manually ordered sections — named groupings that a developer defines and arranges themselves, rather than relying purely on recency or automatic sorting. A developer juggling several parallel projects, for example, can keep each project's threads together in its own section instead of hunting through a single interleaved timeline.
This capability builds directly on the thread-management features Codex CLI has shipped over the previous two releases. Version 0.145.0 introduced experimental paginated thread history with content search and persisted names, and version 0.146.0 (July 29) added thread pinning, forking, and the ability to switch between side conversations without losing the main one. Sections are the next layer on top of that foundation: where pinning surfaces a handful of important threads and forking branches off a single thread, sections let developers impose their own organizational structure across the entire thread list.
Faster Browsing for Long Transcripts
Alongside sections, Codex CLI 0.147.0 introduced incremental browsing for long transcripts. Rather than loading a session's entire history at once, long transcripts now load progressively as a developer scrolls, reducing the wait time and memory overhead that previously came with reviewing lengthy, tool-heavy Codex sessions. This pairs naturally with the paginated thread history introduced in 0.145.0, extending the same performance-minded approach from the thread list itself down into the contents of individual conversations.
Both changes are aimed at developers who run Codex CLI as a daily driver across many concurrent efforts — a workflow that has become increasingly common as Codex has added longer-running agent capabilities, multi-agent orchestration, and remote execution over recent releases. As session counts grow, simple chronological lists stop scaling; sections and incremental loading are Codex's answer to that growing-pains problem.
The update ships via npm install -g @openai/codex@0.147.0.