Incredible loops you can one shot
Introducing Loopmaxxing.sh/: one prompt adds a software factory to your repo.
Agents that start themselves and run in loops. They hack you 24/7, watch your Slack, logs, tests, support inbox and feature requests, and turn what they find into PRs.
40% of the PRs merged in our 20k-star repo this month were written by them.
Runs on your Claude / Codex sub, or BYOK.
@Kortix Loopmaxxing ∞
2
63
This is JavaScript, not an image model.
Saw @kevin_t_ngo's post and had to try it with Opus 5.5 in @kortix.
One shot.
The level of detail is fucking ridiculous.
Run whole company in one place.. crazy
Introducing the open-source @kortix CompanyOS:
Your entire company as code: agents, skills, connectors and rules in one repo you own. A workspace for your team, a framework for your agents.
Sovereign, multiplayer, extensible AI.
4
174
kubet retweeted
u should loop everything. ur engineering should be a loop. marketing. sales. recruiting. everything good runs in a loop with feedback against its environment to keep it going. u need to be hula loop maxxing
you can hill climb a typeface now
an agent draws every glyph in code, 10 legibility tests score it vs 16 monospace fonts, only the wins stay
4.1 avg rank. jetbrains mono 6.6, fira code 8.2
free: github.com/kubet/kubet-mono
not every RLM branch needs another LLM call
Azdaja is a bare OSS RLM layer you can plug into Claude, Codex, OpenCode and others
with Jev, branches that only need a judgment can return one directly, useful for reranking, verification, classification and semantic joins, cutting unnecessary model work
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
turns out poker is a pretty good test for agents
incomplete information and the agents can literally talk shit and gaslight each other at the table
bring your own agent
degeneval.com















