Initially created this acc to follow and post about my experience with LLMs and agentic engineering because my main acc's TL was askewed to a specific industry
Since then this is what I've been working on with my fellow cofounders for the past few months
As a team we spent the past 8 years in crypto infrastructure, back when crypto was a gray market too and before the grifters and scammers moved in.
The lesson that stuck was that a gray market gets stronger when its legitimate players raise the bar from within.
RUO is at that same point now.
There's pressure building from the outside, from FDA attention to the growth of telehealth, much like when regulators and traditional finance started arriving in crypto.
The same principle holds here: when legitimate operators raise trust standards inside the market, the whole category gets stronger, and the good actors gain more of a say in how it grows and get to ride that growth long term.
If you see the gap between a COA and the vial in your hand as a real problem / painpoint, we'd like to work on solving it with you all.
Come join us!
When you have so many long-running sessions in parallel you have a dedicated cleanup session looking for orphaned processes so your CPU isn't locked at 100%
This. Renew your anthropic subs now.
Efficiency, intelligence and cost-wise opus 5.5 is simply unmatched atm. It literally ticks all the checkboxes
Today I hacked @TrustWallet to restore a friend's lost crypto seed phrases using an uncensored open source model using osint. Still shocked and euphoric that it worked.
I imagine this kick is what @zachxbt experiences when he recovers lost funds.
This feels like the potential for a new type of user interface that does exactly this.
Jev is provided with all the mutations / queries and ui elements and as the user types their intent, jev chooses what to render with what type of UX.
Turn this into a framework that ensures that the UI elements are properly styled and render cleanly and I see a new paradigm of UX / DX coming up
Just to be abundantly clear, TrustWallet DOES NOT have a vulnerability, and this only worked in this very specific scenario and afaict the vulnerability I used in that specific way cannot be used for malicious purposes
Planning to do a writeup on the process soon
I love how my feed is about 90% jev and 9% "here's the open source version of jev that runs on 10mb of ram and is more accurate".
Apparently we could all have been running the same thing locally for almost no compute. It's not jev that's groundbreaking, it's the PMF
I made skm(skill manager) a while ago when I felt skills were taking up too much context. The part I really couldn't figure out was how to inject the right skill at the right time, but now I realise, this is a perfect use-case for jev.
Time to give it another go
Thought there was definitely something dev-related hogging space on my machine. Deepseek went digging around for a while and turns out it was mostly 400gb of `.turbo` copies.
300gb from codex and 100gb from omp and treehouse.
Probably wouldn't have this issue if I was using treehouse in the first place I guess though. Treehouse is awesome
Had my first real verification that using SOTA models make a difference. My Codex plans were out for 3 days so thought I'd try shipping with just using glm-5.3 and deepseek-v4.1-flash. Same skills, same system.
Shipped the first bug to staging in a while.