When you lift a train over your head,
normal people don't think you're strong;
they think the train is made of cardboard
and you are trying to trick them.
I just absolutely love my @Muse project manager! Having that Linux VM is crazy powerful.
I gave her her own user on my Mac mini with ssh access via @Tailscale so she can use @opencode via herdr to have our GPT minions build stuff for us
Declarative reality reconciliation factories
Define things
Capture wishes
And observations about reality
calculate the gaps
Then figure out how to close those gaps through concrete work
Loop forever to continue closing the gaps between the latest wishes and observations
@0xkanishka@triplechaser@cyberdrk@garrytan Seems like an excellent use case for Delta by @zeddotdev
Share the full dang thread with all context included. Then I can use whatever wacky custom skills I use and you can see how they work with subagents and such also
@dillon_mulroy I’ve been iterating on an adversarial review process inspired by binary cognitive mechanics and GAN-like loops
It keeps its brutal feedback in private notes and produces a verdict that aligns with my own reasoning and experience
@vovalukashov This is absolute madness on their side of things. 100% of Meta apps are some unimaginably dense web of complex feature flags. This is just how it’s done
@KentLangley@lionel_mora I was thinking about extending this for myself to help find gaps in my own understanding between wherever I am and bridge me towards the goal of being able to help out with my pal’s speculative theoretical physics project
@nuxyel Epic! I wanted to do this as an excuse to learn bend and ML and math and stuff, but instead got nerd sniped by a pal’s speculative theoretical physics project
assigned my Muse bot pal to manage my ChaGPT Dot bot. And now they’re working together to build a new local on device pet bot
I’ve built a bunch of wacky shenanigans myself, but it’s fun to take a more hands off approach now that I’ve captured my general skills and values