cofounder @onyx_terminal
- Tweets 1,993
- Following 574
- Followers 2,071
- Likes 6,140
hyperliquid native lending — trade on L1, borrow on EVM
who is this building this (i am)
13
16
205
21,893
going to start using this the way everyone drops the airplane-survivorship-bias pic and says nothing
10
6
128
8,483
circle’s biggest fumble is how they dropped the ball on hyperevm.
they had the chance to be an enshrined stablecoin for hl and they just… ignored it.
well, now they get to sit on the sidelines and watch tether eat their lunch.
10
5
119
17,032
CoreWriter update: 2 new actions were added which allow order cancellation using precompiles.
With this update, HyperEvm contracts can autonomously set limit orders and cancel them too!
3
8
72
4,600
aave now publicly against those who support usual suspects like morpho, gauntlet and mev cap
not the best look for aci… but don’t blame the player
18
3
70
18,799
psa for hyperliquid teams:
we’ve deployed and verified Bundler3 and GeneralAdapter1 by @MorphoLabs on hyperevm mainnet. feel free to use!
Bundler3: 0xd368763a2aAe21d7f097463aC3CFF79c5FA54522
GeneralAdapter1: 0x2E3E9C4F89A4bb7D3ca345EB5F4BaD1555f32E3E
3
6
63
9,268
i don’t think about hyperevm as a standalone chain anymore. it’s a layer to store arbitrary state and executable logic on hypercore
3
2
59
4,839
in a reductionist sense:
- curve’s edge is tech
- aave’s edge is risk mgmt
- mkr’s edge is scale
as a result, always pay attention when
- curve releases new tech or sunsets products
- aave changes risk params and market configs
- mkr implements dai allocations and dsr changes
1
7
59
3,014
Mor precompiles! Some new HIP3-aware Read precompiles that flew under the radar:
AccountMarginSummary: Get the account value, margin usage, and aggregate position info for a user
Token Supply: Get details about a token’s max supply, total supply, circ supply, emissions etc.
BBO: Query top of book bid/ask prices for any pair
CoreUserExists: Check if a user is activated on HyperCore. Very helpful when estimating fees and complex interaction costs.
These flew under the radar, probably because they aren’t displayed as prominently as CoreWriter updates on the docs, but they’re very important nonetheless.
Notably these precompiles are HIP-3 aware: It’s easy to see they’re built with multiple perp dexes in mind. More people need to be talking about this.
HIP-3 might be closer than you think.
5
4
57
7,429
it started with a simple question: can users get more leverage than hyperliquid, on hyperliquid?
yes they can.
2
2
53
4,402
in hindsight there were three ways to price USD0++
1. fixed price oracle: fast setup, not market-priced, some risks
2. pyth: slow setup, market-priced, trust assumptions with pyth
3. custom oracle: slowest setup (need audits), custom pricing, custom risk
morpho chose (1) to be early. they got a ton of tvl and market share. unfortunately they had to deal with market risks. small L but feels like they’ll come out fine.
eul chose (2) and might have a smaller market share, but their systems handled volatility better and they have fewer people shouting at them today. w for them.
meanwhile aave is having a field day because they never touched usual.
not much to say — it’s just tradeoffs all the way down
11
2
53
7,743
i’m conflicted on this pov but rough thoughts:
- forks are battle tested code, yes
- there is no way to fork risk and team exp
- projects that build from scratch are high risk reward bets on novel stuff. imo that’s very important
- it is dumb to judge a chain by the prevalence of forks, esp this early
in this particular case - most people saying “hyperevm is just forks” are either trying to fud hyperevm or don’t understand it. ct can’t deal with nuance.
“HyperEVM is just forks”
Yes, thank god we have battle tested code to borrow against $12-13B worth of deflationary collateral
8
1
44
6,084
new write precompile spec just went live. great changes overall.
tldr:
- the contract is now called CoreWriter
- resting orders are here!
- upgradeability + backward compatibility was top of mind for the refresh
- precompiles need to be flexible and dynamic, but also granular and lego-like. this update does exactly that.
- there’s no way to predict what precompiles are needed in the future, this change has upgradeability built in
- it might be slightly tougher to debug txns with the new spec, due to nested encodings. imo its worth the trouble.
- new devs might struggle with encoding actions and testing simulated actions across hypercore / hyperevm. skill issue tbh.
- looks like the team is actively looking for feedback on new precompiles to be added, i think we’ll see semi-regular updates and additions to list of actions
hyperliquid
3
5
43
3,101
i think sentiment did loops way better than others:
- higher leverage (10x vs 3x)
- no big blocks needed (significantly cheaper)
- works with native hype (better ux)
- live since day1 (shipped months faster)
credit where credit is due though:
- others have a lot more liquidity
- their marketing was better / costlier
- ui is normie friendly / more informative
i’m taking notes. most of this is fixable for us, but this is also me saying — check out lev loops on sentiment. it has more strats, and is way cheaper, with a better ux.
as always, looking for feedback too. feel free to dm me on tg (@\ruvaag) with thoughts or questions.
7
3
40
2,875
our new perps x lending product is internally called janus — after the roman god of “gates, transitions, and duality”
it reminds me of hypercore x hyperevm interactions and precompiles
soon
2
3
37
2,133
“they call it ethereum joe, ok? and look, it’s old, it’s slow, it’s expensive. but once i’m president i’ll fix it. i’ll fire gary and ask every etf to stake their eth. and you know i love tariffs joe, i’ll put tariffs on all infra and L2s. it’ll be great joe, i’ll fix eth.”
3
6
38
3,400
the most hyperliquid-aligned thing a project can do is build on hyperliquid. thank you for your attention to this matter!
2
2
30
5,659
did i miss the memo or were two new methods added to corewriter recently?
idk what ‘finalize evm contract’ does but ‘add api wallet’ is huge.
how is no one is talking about this.
8
1
31
3,432
maker is opinionated about using only lindy assets with conservative ltvs as collateral. they can make assumptions about expected volatility and allow delayed liqs.
new markets want to support all types of assets + a range of ltvs. they can’t make the same assumptions.
1
7
367
$30m in borrows - huge milestone considering sentiment never had >$5m in supply side liquidity
this means the entire supply tvl was lent out more than 6 times over
capital efficient liquidity.
3
1
30
5,388
next time i write contracts i will
- only spend time on things that no one else has implemented
- fork heavily-audited code wherever possible (with attribution)
- never use inheritance
take advantage of past audits
don’t reinvent the wheel
and lastly, inheritance was a mistake
4
29
1,407
new lending narrative brewing? euler evc, morpho blue and astaria starport are all building a "seaport for lending"
looking forward to how things play out as they launch and scale
6
2
27
5,263
sentiment v2 contracts are finally public!
some highlights:
- improved risk mgmt for lenders
- rehypothecation galore
- complex flows reduced to one-click
- safer contract architecture
hmu if you have thoughts or wanna talk more, i'd love to hear from you
1
5
28
1,688
first thoughts on morpho v2: they’re going hard on the defi mullet + curator models. the whole system is designed with the intent (!) to onboard more curators including corpo / tradfi ones.
noticeably, this is very diff from aave v4 which doubles down on the monolithic design.
2
1
25
1,904
Max is the only one on my feed who talks about HL L2s / app-chains with conviction. Maybe because others have ptsd from past cycles. He’s right though, the opp is huge.
Caveat: I don’t think HL L2s will be positioned as such. They’ll look like super apps or specialised infra.
2
1
26
2,852
why doesn’t the “cracked” swe have an active personal blog?
he can’t find time to build a static site generator from scratch in C
these Smart Software Engineers are so allergic to copying code that they'll spend weeks/months on complicated solutions that are impossible to maintain because they didn't want to do something simple/dumb. Something a "non technical" person can do
2
24
3,314
a sentiment liq bot took the first flash-loan on hyperlend / hyperevm? that’s lore.
The first flash-loan on HyperLend (I'm geeked af for this)
> borrows 3.00035 wHYPE via flash-loan from HyperLend
> liquidates a position on Sentiment (HYPE borrow and wstHYPE as collateral)
> swapped wstHYPE back into wHYPE using KittenSwap
> repaid the borrowed wHYPE to HyperLend
> Total Profit: 0.97 $HYPE ($12)
When flash-loan liquidations through the L1 orderbook?
2
2
25
2,499
reading the new @BaselineMarkets whitepaper and love to see how much thought they’ve put into every aspect of the curve. bullish!
2
5
25
3,231
devs are printing with internal tests. how would you like to get more leverage than hyperliquid, on hyperliquid?
coming soon to a (new) sentiment product near you.
2
2
24
3,797
here’s an idea—we extend this arch to collateralise perps on hl, traders can offset funding with farm yields
something like:
> deposit 100k USDC as margin
> open a 3x btc position on hl
> borrow 80k usdc on the evm against the same 100k
> farm using 80k to offset funding costs
5
7
22
4,456
it's interesting how aave has billions in sticky tvl earning low yields
makes you wonder if LPs:
- believe aave has the highest risk-adjusted apr
- don't care about lending apr because they can outperform using only borrowed assets
- possess super strong consumer inertia
5
21
2,209
DL is the best crypto team when it comes to "understanding what your user needs" — and shipping
i've been a llamasearch enjoyooor
since day 1, and can already see myself using this everyday. great work!
new llama launch!
we built a new homepage experience with all the stuff you might need, plus you can customize it
2
23
5,443
amazing new stuff from curve. seems like their response dsr and gho staking, but maybe even better
i see this as a huge net win because it meaningfully allows people with size and institutional risk mandates to gain core defi exposure
would love to read more about the off-chain stuff though
By staking your stablecoins on stablecoin platforms such as Curve, you’re effectively giving a loan to Curve the protocol.
Curve controls monetary policy via fee splitter and decides if it should increase or decrease the yield of crvusd.
It’s like tokenised t-bills but then decentralised and fully on-chain, and scales with demand.
@CurveFinance effectively introduces a new decentralised market control to govern the stablecoin, by using a parametrised fee splitter module that siphons dynamically and algorithmically adjusted portions of real revenue and funnels it for growth.
This is a scaling strategy. Today we do this for stablecoins, tomorrow we hope to do this to fund other endeavors that may or may not be on-chain!
2
3
22
2,394
re khype: some posts mention a wsthype stress test
but very few talk about the second-order effects:
- dex liq fragments + reduces
- dex aggregators issues (?)
- low liq on lending markets
it can turn into a stress test for everyone. don’t let the bots get your hype on a wick
1
20
2,549
7/
@sentimentxyz is a lending protocol that aims to utilise perpetual positions on @HyperliquidX as collateral via isolated borrowing accounts. This unlocks potential opportunities surrounding farming strategies, spot-perpetual pair trades, and more.
@liminalmoney introduces xTokens, which tracks various delta-neutral strategies. This allows users to earn yield on their held assets while hedging their exposure to market movements.
1
1
21
1,342
defi has been exceptionally great at transforming everything into a liquidity problem
not much else to care about other than liquidity - acquiring it, managing it, and having too much or too little of it
and everything else tends to fall into place with a robust liquidity base
1
19
3,204
i just know if he was around today, he'd own the algo with his hot takes in a way no one can imagine
born a few generations too soon
18
399
congrats — i loved reading the starport paper and followed astaria on base. need more novel attempts to long-tail lending + focus on intents.
1
3
499
people dunk on nihilism as a philosophy — but it is a rite of passage, everyone is lured by it at some point
the only difference is some gaze into the abyss long enough to see through it
20
2,780
usdt borrow rates against wstHYPE (and pool liquidity) on hyperevm right now:
sentiment: 3.26% (~700k)
hypurrfi: 8.54% (~1.7m)
felix: 8.96% (~4.8m)
hyperlend: 4.94% (no more liq)
i spot some cheap borrows and rate arbs here.
4
19
1,369
write precompiles coming soon!
CoreWriter (fka L1Write) was just included to mainnet bug bounty payout scope.
1
17
1,268
my current pov on hypercore oracles vs standard oracles:
hypercore oracles are more secure than standard oracles because they are built directly into the chain as a precompile — they provide price data from the node itself. they function reliably as long as the chain is up with the only risk being the critical node infra.
standard oracles, however, add their own infra and sc risk on top of the risks that hypercore oracles have. in that sense they’re inherently more risky.
i think, over time, all hyperevm price sources will default to hypercore oracles
this is a great article and i’ve talked to eduardo about it separately since my views differ a bit.
i think hypercore oracles are safer in terms of availability and safety because they don’t stack risks. imo they should be the standard solution.
the real question for me is whether hypercore oracles report prices just as accurately or not. this can now be checked with a retro since its been a while.
fyi as of today sentiment uses standard oracles. we’ve deployed contracts for hypercore oracles but we don’t use them yet. just testing.
5
18
2,045
this turned out extremely well. i designed the initial system completely based off hyperliquid and other known defi ux legos
if i have to explain to a user how it actually works — that means the ux isn’t good enough
when people talk about using perps as collat, they shoehorn it into an aave-like flow where they “deposit” the perp as collateral and borrow against it, as if it’s a token.
that sounds like a terrible ux to me. i like to nerd out about it, but if i have to explain the inner workings to an end user… sudoku.
the product should completely abstract all that away, and it will.
1
1
18
1,316
what an unforgettable last week, but now we can finally go back to building - safer, better and even bigger than before
We are pleased to announce the Sentiment protocol has been restored to its full capacity. Users can now create accounts, lend, borrow, trade, and withdraw.
We're back at arbitrum.sentiment.xyz/ 🌀
1
17
1,099
some lore: sentiment v1 was the first ever audit contest on sherlock
i still remember when the format was so new that it was ‘pitched’ to us by @jack__sanford — obviously, we loved it. no-brainer.
since then we’ve used sherlock’s insurance, had multiple contests and will continue to do so!
more recently, sherlock’s new bbp and revamped judging are also steps in the right direction.
crypto security is a different beast and it only makes sense to learn from the past, find novel solutions and keep improving our practices!
this might seem weird today, but there was a time when audit contests were considered ‘experimental’ and ‘novel’
their efficacy was debated against private engagements as an either/or argument
ofc we now know that both formats have their own place in the security stack
1
3
17
6,234
rm woke up to a shoutout in the rust x eth day talk by @msolomon44
some context to steelman his argument:
- my first rust pr in prod
- foundry dev chat was very responsive (thanks @hievalir)
- only took a few hours to go from first reading the issue to having my pr merged
2
17
781
the trading ux on sentiment is about to get a huge boost over the coming days:
- faster refresh rates
- routing using aggregators
- clear info about output amt, price impact, slippage, etc
better execution + improved ux + cheaper rates = more $$$ for traders
1
1
16
806
> paper about lending markets
> published last month
> skims the intro
> “main lending platforms, aave and compound”
> compound ???
> no mention of morpho or euler
> i’ve read enough
> close tab
2
15
642
apparently morpho's max withdraw feature puts the position right at the liquidation threshold
while this is what "max withdraw" truly means, it makes for poor ux. a position at the liquidation threshold can be liquidated on the slightest of price movements
most interfaces implement max withdraw with a ~5% buffer. the easiest next step for morpho is to put in a similar fix
fwiw op is completely wrong for calling them scammers. it's a bad look for a mature team like morpho, but def not a scam
3
1
15
2,371
and i would have gotten away with it too, if it weren’t for those meddling objectives and path-dependencies…
14
399
current hype lending rates:
felix - 0.26%
hyperlend - 0.35%
hypurrfi - 0.92%
sentiment - 3.99%
good time to rebalance your supplied liquidity.
1
1
14
812
i agree that this is not the dao’s stance today
but to call aci ‘one of many delegates’ greatly understates its influence
15
1,175
i think we'll see some innovative products built on seaport just as we did with uni v3 conc liq
seaport is more than just a "backend for opensea" or a "nft marketplace protocol"
it's closer to a generalised onchain exchange and orderflow primitive
1
14
869
fascinating how people keep bashing ohm because of the initial dump—it’s one of the best projects that played out long term
whoever brought below or close to backing is up a lot, cooler good, no crazy drawdowns… idk what’s not to like
1
13
382
grug’s pov is accurate + pragmatic.
literally nothing apart from btc (and maybe eth) has reached a scale where decentralisation matters.
decentralisation is elimination of trust assumptions, but if everyone involved trusts each other (as with hl) — it’s redundant
2
14
1,817
something i wish i'd internalized way sooner ––
there are no under-collateralized loans. they simply don't exist. you're always collateralizing with assets, upnl, future flows, or something non-tangible (credit score, trust, IP, etc.)
this was a doubly important learning because we initially positioned sentiment as an undercollat lending platform. so glad we course corrected asap.
1
12
679
some facts from dl:
- the only lending protocol with any meaningful revenue is aave with ~40b tvl
- the second, morpho has 6.75b and no revenue
- tvl / rev is even smaller further down
tldr everyone in lending is focused on growth and that is the best use for revenue (if any)
Question for all the lending protocols building on HyperEVM
What are you currently using your revenue for?
Avoiding this question will result in me guessing you're not aligned with HL ethos 😇 Salute
@hyperlendx @felixprotocol @HypurrFi @hyperdrivedefi @sentimentxyz
1
1
11
1,117
the elites have never seen a reality that doesn't rely on rent-seeking systems or paternalistic governance
this is just their ancestral wisdom being applied to a modern context
1
9
400
it's a great time to ask people about their crypto thesis with an invalidation
personally, my thesis revolves around crypto driving efficiency and it's only gotten stronger through the bear
price down but conviction up
1
11
2,271
what’s the best way to explain the symbiotic relationship between hypercore vs hyperevm?
i need something better than “imagine two chains running in parallel with overlapping state, connected by precompiles”
4
13
994
rareskills needs to repurpose their 'solidity interview questions' into this. underrated resource
why do we need leetcode for solidity though?
1
13
428