Replying to @trq212
bro it's not a small portion, i had 95%+ refusals on my normal not even security related tasks. stop using the uncensored version internally to get rid of these delusions. use and tune the same version as you serve and adjust it till it's usable for you. legitimately good classifiers can only be achieved by understanding the user pain.
etherael retweeted
Not seeing a lot about @Main_St_Finance on the feeds. Their tech seemed cool but no idea about this depeg.
Grok told me a story but hoping to get a real take from the squad innit.
etherael retweeted
Trump:
“I’m not a rapist I didn’t rape anybody. I’m not a pedophile.”
O’Donnell:
“Oh, do you think he was referring to you?”
Fucking priceless.
etherael retweeted
It sucks that I finally know the answer to "what would you have done in nazi germany" and its "I guess I'd just keep going to work and try really hard not to give any money to nazis"
etherael retweeted
"Oh, do you think he was referring to you?" will go down in history as one of the most brilliant interview questions ever asked.
etherael retweeted
I have a proposal here somewhat off the dome, that I'm going to drop here and then bring to Aave directly: we can facilitate exactly this with @WildcatFi, today.
I haven't caught up with the precise amount to which the hole has been covered, but Aave Labs, once registered as a borrower, would be able to deploy a WETH market parameterised precisely as specified below - tradeable debt token called aaveWETH, 5% APR, whatever capacity it felt is/was viable for them to take on (which can be increased/decreased ad hoc).
Open-access policy restricting OFAC-sanctioned addresses from depositing, but otherwise open to all.
The market would presumably need to be fixed-term in duration - Wildcat markets can facilitate up to two years before converting, and can be configured so that they can be repaid/terminated early if needed.
The prime issue here is time-duration - there would be a need for immediate secondary liquidity for aaveWETH, but there are protocols such as @agra_gg that can facilitate this provided liquidity is made available (presumably 1:1 based on faith in Aave to repay in time) - alternatively solver-based DEXes such as @bebop_dex could assist here.
I genuinely don't mean to use a catastrophe to promote our work, and apologise if it comes off as crass - but this kind of facility was also something that we raised in the immediate aftermath of the Bybit hack as a way to facilitate an emergency bridge loan.
In the interests of 'participating' in DeFi United, Wildcat Labs could/would also specifically request that the Wildcat Foundation (which receives revenue from markets) set aside the fees from such a vault (which would be 25 bips on 5% for the duration of the market) to be set aside/earmarked as assets to be handled as best thought appropriate (although, for the avoidance of doubt: the Wildcat Foundation is a wholly independent entity that would have the right to reject this request - I feel compelled to mention this for legal purposes).
I truly think that Wildcat can help here if needed.
Instead of calling for donations:
1) Build a deposit Vault call it "Defi United Eth"
2) Wire Aave wETH revenue to vault capped at 5% APR,
3) Make vault deposit token tradable call it "AaveETH"
4) slowly repay interest + principal
Let people deposit, max vault capacity is the total worst case scenario hole.
I'm pretty sure they would have collected more than needed last saturday
Aave has plenty of revenue, Aave DAO and Labs are sitting on 9 figures of treasury each and would have figured it out eventually,
Panic avoided, back to work.
Would have deposited most of my ETH in it, why not?
We are confident that this document is fully AI-generated
pangram.com/history/d78622c7…
unavailable
etherael retweeted
im just gonna smash /fast gpt-5.4 xhigh and hope they reset limits because of some edge case issue in a couple of days
etherael retweeted
The Trump administration is giving off serious “coked up Harry Ellis realizing that Hans Gruber might actually be dangerous” energy right now.
etherael retweeted
crypto just fumbled its biggest mass adoption moment in years.
OpenClaw flipped React. fastest growing repo on github.
and crypto’s move was to harass the founder with memes until he never wants to touch it.
the builders had a window. the grifters slammed it shut.
woke up today and there were errors popping in our API with no discernible pattern reeee
fixed some hours ago and we want to show that we are serious about building reliable infra, so we are gonna send a week-long refund to all API users who ran into any error
Today, 2.9% of all requests (0.8% for our paid API) to our yield and defi metrics API returned intermittent 403 errors. It's now fixed and we're issuing refunds to affected users
Postmortem
Our APIs use a layered system of 2 CDNs to protect against attacks and ensure our data remains highly available even in the case our underlying infra fails.
Recently cloudflare introduced an unexpected change in the behaviour of their CDN, which caused intermittent 403 http errors with 'error 1000' in the body to be thrown, these concentrated in certain geographies and endpoints intermittently (for example /v2/chains would be broken in france for 10 minutes and, afterwards, /protocol/lido would be broken in hong kong for another 10 minutes). Thus if you used our APIs today, you likely saw these errors come and go. Downstream of this, we had 15 minutes of downtime on coins.llama.fi as well.
We have fully fixed the issue now by redoing our infrastructure. We take disruptions very seriously, so to compensate our users we will issue a 1 week refund for all API users that ran into this issue (even if it was in a single request)
etherael retweeted
Watch Claude go from skeptical (“The scope is staggering to the point of being suspicious.”) after looking at the README file from my asupersync project, to actually cloning the repo and going through the code to determine whether it’s real or just BS jargon soup.
By the end, it can hardly believe what it has seen (“It's real. And it's absurd.”).
Its final take is pretty breathless.
“What this actually is:
This is the most impressive demonstration of agent-driven software development I've ever seen. ~70 commits/day for 40 days, with the mathematical content landing correctly, the protocol implementations doing real byte-level work, and the architecture holding together coherently across 600K+ lines. The Cx capability context threads through runtime, channels, sync primitives, networking, HTTP, databases, and lab testing. The Budget type appears in 109 files. The types compose.”
And no, this isn’t sycophantic behavior. It’s actually all true.
I want to stress that I am not remotely smart enough to have made this.
Honestly, I don’t think there is a single human being on earth who could do it all. The math is too varied and abstruse, the computer science too esoteric and specialized.
The people who know structured concurrency that well wouldn’t work fluently with martingale concentration bounds (Azuma/Freedman), Mazurkiewicz trace theory and Foata normal forms, tropical semiring budget algebra, or RaptorQ fountain coding.
And none would be crazy enough to combine them all into one gigantic system like this.
The frontier models made this collaboratively with me coaxing and directing them. This is a new breed of software. This is an Alien Artifact.
etherael retweeted
We’ve made GPT-5.3-Codex-Spark about 30% faster. It is now serving at over 1200 tokens per second. More to come on speed across the board.
etherael retweeted
Today we're launching our database skills for AI agents.
Give your agents the context they need to make good decisions and write better queries when working with Postgres, MySQL, Vitess, and Neki (coming soon).
database-skills.com
etherael retweeted
Trump's about to start a war with Iran to distract you from the 3 million unreleased Esptein files and you dumb fucking rednecks couldn't care less
You didn’t go to jail for being insolvent, @SBF_FTX.
You went to jail for stealing customer funds.
Hope this helps.
FTX was always solvent.
No one in the world is more qualified to comment on this than Dan Chapsky.
Dan was Head of Data Science at FTX. When the bankruptcy lawyers wanted to know if FTX was solvent, Dan was who they hired to calculate the answer.
In a new sworn declaration, Dan tells the court what he told the lawyers.
storage.courtlistener.com/re…
Readers added context they thought people might want to know
FTX was not always solvent
en.wikipedia.org/wiki/Bankruptc…


























