Many people get stuck when working with AI, not because they write prompts poorly.


It's because the same conversation gets more and more bloated over time, eventually turning into a garbage dump of context.
I’ve now added an "auto shell change reminder" to my Codex workspace.
I have 10 pinned workspaces every day:
Content, research, strategy, product, daily wrap-up...
/morning, /next, session-end tasks are all stored there.
The problem is, after using it for a while, each thread's historical baggage gets heavier and heavier.
Tokens keep piling up,
Old tasks, new tasks, wrap-up info all mixed together.
In the end, the model isn’t not smart; it’s that the workspace needs a shell change.
So I let automation monitor tokens_used:
Exceeding 50M
→ Automatically scan pinned threads
→ Remind about Daily Rhythm rollover
→ Archive old threads
→ Hand over to new threads
This may sound like OCD, but it’s actually the hygiene system for the AI workflow.
Someone who uses Codex / Claude for real business over the long term, in the end, it’s not about whether they can write prompts, but whether they can keep the workspace maintainable.
View Original
post-image
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned