_discochuck_ (🔺doption arc ) retweeted
A few hours ago, a USDC tranche vault on Avalanche lost ~$1M due to inconsistent principal accounting in the fund-user claim path.
The vault has two roles: the fund provider, who deposits and withdraws principal, and the fund user, who borrows funds from the tranche and is responsible for repaying them. The attacker controlled both roles within the same tranche, allowing them to deposit funds as the provider and immediately borrow them as the user.
The claim function deducted borrowed but unmatured principal from total deposits, but ignored matured yet unrepaid principal. When the attacker claimed again after maturity, the function first processed maturity records using the current timestamp, moving the previously borrowed principal into the matured balance, then checked the available amount. Because the calculation omitted the matured balance, it incorrectly restored the full borrowing allowance, even though nothing had been repaid.
Flawed formula:
Available = Total deposited − Borrowed, not yet matured
Correct formula:
Available = Total deposited − Borrowed, not yet matured − Matured, not yet repaid
The fund-provider withdrawal function correctly deducts both balances, while the fund-user claim function omits matured but unrepaid principal.
Key attack steps:
1. The attacker created a tranche with the same address as both fund provider and fund user, setting its maturity shortly afterward.
2. Using an Aave flash loan, the attacker deposited 1,013,353.61 USDC as the fund provider, then borrowed the entire principal as the fund user in the same transaction. They used those funds to repay the flash loan, paying a 506.68 USDC fee.
3. After 163 blocks, once maturity had passed, the attacker called claim again. The incorrectly restored allowance let them borrow another 1,013,353.61 USDC without repaying the first tranche loan or making a new deposit.
The second borrowing was funded by other users’ principal deposits in the vault.
Attack TX: app.blocksec.com/phalcon/exp…
_discochuck_ (🔺doption arc ) retweeted
Most crypto startups get their first data hire wrong
They wait until they raise, run 30 interviews, and hire a junior analyst because it feels like the cheaper call. Six months later that person is still figuring out where the data lives and nobody has decided which numbers actually matter
Your first data person has to decide what's worth measuring, not just write queries. That's a senior job, even if it's part-time at first
Founders, who was your first data hire and would you make the same call again?
_discochuck_ (🔺doption arc ) retweeted
Replying to @Red_Pill_US @grok
The only rational explanation is that Kanye was Dutch intelligence this whole time.
_discochuck_ (🔺doption arc ) retweeted
Median pay on verified crypto data job postings right now:
Data analyst: $126k
Data engineer: $196k
Data scientist: $203.5k
That's a ~$70k jump from analyst to engineer. If you already write SQL all day, picking up pipelines (dbt, Airflow, Python) is probably the best-paid upgrade in crypto data
If you're an analyst, what's stopping you from learning those new skills?
hashcat v8.0.0 released!
The largest update in the history of hashcat, with PCFG attacks, hash recipes that crack hashes without a hash-mode, and bcrypt on FPGA boards.
Release notes: github.com/hashcat/hashcat/b…
Download: hashcat.net/hashcat/
_discochuck_ (🔺doption arc ) retweeted
Playing my new Tower Defense game, Angels of Doom for @KennyAVAX and it's easily in my top 3 I've created.
My first data job was making maps for an insurance company in Paris. Geography degree, no CS background, zero crypto
A few years later I was running BI at Binance, then became Morpho's first data lead
So when founders tell me they'll only hire someone with 5+ years of onchain experience, I kind of laugh. Some of the best crypto people come from weird places, you just have to know what to look for
air gap
smh
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
_discochuck_ (🔺doption arc ) retweeted
Dick & Tardy: Problem Solved
Episode 01: Ribbon Cutting
enterthegrotto.xyz/games/gam…
Just launched on @TheGrottoL1 - Let's game!
🤖 Made with AI
_discochuck_ (🔺doption arc ) retweeted
$phar is doing the avax special rn
no resistance on the way up
no support on the way down
a dca like @TradingAloha is doing probably plays out nicely if you’ve got the patience for it
happy hunting
_discochuck_ (🔺doption arc ) retweeted
The Bog event runs all through Halloween for free in Bob's Petroleum only on @TheGrottoL1.
Spend spooky season with us.
_discochuck_ (🔺doption arc ) retweeted
Pieces of Halley's Comet are going to burn up over most of the country before dawn Thursday.
If you're in the clear, step outside after the moon sets, around 3 to 4 AM, and look up. From a dark spot, 10 to 20 an hour.
_discochuck_ (🔺doption arc ) retweeted
Coinbase, Robinhood and Kraken are hiring data people!
Shoot your shot and apply now:
- Senior Machine Learning Engineer, CX, @coinbase , Remote US ($191k–$225k) → unchaindata.xyz/job/coinbase…
- Staff Data Scientist, People Analytics, @RobinhoodApp , Hybrid US ($217k–$255k) → unchaindata.xyz/job/robinhoo…
- Senior Analytics Engineer, Finance, @krakenfx , Remote US → unchaindata.xyz/job/kraken-s…
- Senior Business Operations Analyst, @chainalysis , Remote → unchaindata.xyz/job/chainaly…
- Business AI Process Automation Analyst, Crypto.com, Hong Kong → unchaindata.xyz/job/crypto-c…
- Senior Pricing Analyst, @Figure, New York ($68k–$102k) → unchaindata.xyz/job/figure-s…
Repost to spread the word! ♻️
PS: DMs are open if you want your job featured on Unchain Data's job board
PS 2: Want to ask about these roles, salaries or interviews? Join Crypto Data Wizards 🧙 → t.me/CryptoDataWizards
--
unchaindata.xyz/job/coinbase…
_discochuck_ (🔺doption arc ) retweeted
A new game comes (late) with a tribute to our dear friend @KennyAVAX who's going through some medical shit.
To lighten the load and do some fundraising for our friend, here's my Tower Defense style game "Angels of Doom: A Tribute to Just Kenny" with a tip function built in.
_discochuck_ (🔺doption arc ) retweeted
Hiring the right crypto data person is weirdly hard
You need SQL + onchain + product + business sense, and most CVs only have one of those
I run the first crypto data job board so I see both sides every day, which is why I started helping teams fill these roles. Just closed one and the candidate starts soon
Hiring an analyst, engineer or data lead? DM me or check the link below























