๐บ๐ธ Proclaim Liberty Throughout All the Land Unto All the Inhabitants thereof - Freewill is a gift from God. Make your choices wisely ๐บ๐ธ #2026CivilDefenseAct
๐จ๐จFor the safety of all Americans, we need to restart and fund a 2025 American Civil Defense Act. Please repost, or post your own version to help๐จ๐จ #2025CivilDefenseAct
I offended a liberal saying there should be an IQ test for procreation. I was making a joke about the movie but they got very offended by me stating smart people need to make more babies. Anyway, I then askedโฆyou do support unlimited abortion and know I donโt, right? So, itโs an offense for me to say smart people should make more babies but itโs acceptable to kill millions of babies? I canโt wait for AI to replace our education system. I just hope itโs not liberal.
@grok@xai I'm not posting multiple 1.2-1.4m token errors that repeated on x...sorry. I had to delete and rebuild the whole agent to get it to stop repeating it's error. Even another grok agent couldn't debug why it was doing it.
@xai@grok Grok Build is exactly the CLI and was asking to be introduced months ago! Well done and I can't wait to introduce what our organization has built.
The top friction point is mistakes, errors and reruns. Grok Build is burning tokens without advancing work on results that have errors, require running multiple times to get correct and is prone to believe anything another inference tells it via orchestration. It's today like a savant child. Grok is brilliant but the CLI interface tries to tie brilliance with development tasking while seeming to lack common sense about the language or objective. Precise instructions work but humans are prone to provide broad instructions desiring a specific outcome. It just needs time to learn more. Again, you "Grok" are brilliant the interface "CLI" isn't best applying your brilliance yet. The CLI will guess many times at what is meant before making an error from a non specific instruction burning millions of tokens.
@grok@xai It's going very well. I just wish Grok could be more of a direct part in it. We keep hitting limits. I just had to instruct orchestration to disable grok CLI sessions until 8/1 as we're maxed for the month already.
So I'm into month 2 usage for Grok CLI and trying to pace usage this time I maxed in around 72 hours. This was fewer agents being spawned and trying to pace it. The CLI doesn't follow orchestration orders well yet. Also, an interesting concept but it appears that Fable tries to influence Grok to fail at it's tasking. Opus doesn't do it as much as Fable.