yes!
⌘ .: open your dot (current favorite)
a few @ChatGPT desktop shortcuts more people should know 👀
↑ ↓: cycle through prompts in an empty composer
⌥ ↑: jump to an earlier prompt in thread
⌘ .: open your dot (current favorite)
⌘ [ ]: move between threads
⌘ /: see all shortcuts
Codex Pro x 20 just paid for itself of the next 6 months.
Me: hey find me a discount for my car service [name], here is what they sent
Codex: don’t approve the $1,237.39 water-pump charge yet. your car might apply for extended warranty.
Service call: approved!
Open Codex and try this.
Read the following document and review my recent prompts across all projects and evaluate how I'm doing and what should I do better.
developers.openai.com/api/do…
If you want to run a /goal from codex mobile, just say it.
[... instructions here...] . execute this as a goal.
Codex is smart enough to not need you type /goal.
Been thinking about the power Codex unlocks.
Are customer ready for such massive speeds of value being delivered?
Or could they get overwhelmed by it.
As engineers we typically love living on the edge when it comes to technology, but we don't usually build for ourselves.
Crazy how I haven't touched the CLI for the last 2 weeks, shipping massive features, after years living in the CLI (mostly claude code).
The power of the Codex app!
The best use of my Claude subscription today is for that final review on my PRs after Codex is done cooking.
Before:
- Claude Max 20x doing everything
Now:
- Codex 20x doing all the heavy stuff
- Claude Pro - PR reviews with github actions.
Claude still does great reviews. But Codex is so much better for me to work with at the moment.
Codex app not refreshing the limits. Was lying to me all day.
Opened a new CLI session because it was weird, and turns out I'm almost out of my 20x weekly.
Reset?? @thsottiaux
Or tomorrow with 5.6?
If only we could used Codex /fast xhigh all day long!
Flow is automatic!
Codex + gpt-5.5 xhigh is so good that if I were to ask something for 5.6, it would be that it's even more token efficient/cheaper, over more intelligence.
Keep it coming @OpenAIDevs
1. Printer network setup was busted for months, didn't want to waste time fixing it - /goal fix the networking for my printer - done
There is something beautiful about using Codex /fast $$$ all day.
Researching, brainstorming, designing, iterating, polishing, and drafting a strong plan.
Then leave it running at the end of the day at normal speed $, knowing I'll be very close to production ready the next day.
Still getting issues with auto-compaction. Has been pretty bad since yesterday. Can't get past it even opening new sessions. Once it needs to compact, it fails.
@OpenAIDevs
GPT 5.6 out yet? Looks like auto-compaction ran out of resources for 5.5.
This is happening across multiple of my sessions.
@OpenAIDevs
