@jelanifuel Did it reset the week's end of fill up the bar? Like, if it Was Going to reset Tuesday and yoy click it sunday.... does it reset again Tuesday or did you reset your week to Sunday?
@rezoundous What happens if you buy a reset and Tibo presses the button?
If I buy a reset, does it reset the clock or refill the bar?
These 2 questions will prevent me from buying now.
@thsottiaux Ability to auto archive threads. Have codex close subagents when not in use. (A multi hour task leaves 100 subagets idle but awake) Can I easily reference thread A inside Thread B? Can I have them monitor a thread for changes?
@thsottiaux You released the pets feature the week after my dog died. I was able to let Austin live on through codex. It even recreated the "tricks" she would do and uses them contextually to communicate codex's working status.
@thsottiaux Max plan = $500 = Unlimited access. No quota limits. (Limit consecutive threads maybe?)
More reliable and faster remote connections from Android to Windows. (Especially with long threads)
Dear @OpenAI 5.6 Sol with Work mode is really amazing.
What would be really cool is if it was also giving a more comprehensive like "operations dashboard" as it was working. Like a whole sidecar visualization I can bring up.
Like, clicking on the subagents is cool and all but you could easily spin up some dynamic UI or something tracking the work items and showing what they are succeeding at and failing at. Like even a sidescroll burndown or Gantt chart type thing would be really cool.
And this request is not purely aesthetic, it would be really helpful to watch what it's doing as it's going so I can be caught up by the time it's done. I'm thinking little green checkmarks for things that worked, red X's for the things that didn't work and required retries, yellow flags for insights and direction changes, that sort of thing.
I took the liberty of asking ChatGPT to mock up what I'm talking about and it pretty much nailed it.
I don't really care to read exact thought traces anymore, but if you could make the successes, failures, insights, etc, explorable so I can learn as it's going. that would be really cool. Particularly if, on a code related thing, if it could highlight the code snippet and insight or something when I drill all the way down.
Or if it's more research related.
Because as people say, learning what didn't work (and why) is often just as valuable if not more valuable than just knowing what worked.