Pinned Tweet
Did i lock into my dev arc and cook something up for my first ever hackathon?
well, tada, here we are...
This is my entry for the @Bitget_AI #BitgetHackathon season 2
This whole idea came from the problems i have encountered with the majority of the "AI trading bots" i have used in the past 2 years of my trading experience...
Most AI trading tools follow the same script: take an LLM, hand it a price chart, ask "buy or sell?", and let it decide how much capital to risk....sounds smart, right?
Sadly, this only It works fine in a demo. It falls apart the moment the market does something the model hasn't seen before.
That's because a single prompt has no adversary, no hard limits, and no memory of its own mistakes. It agrees with whatever narrative it's fed, hallucinates conviction during a flash crash, and worst of all, it is often the same system deciding both the trade idea and the position size....now you are left standing between a confident-sounding hallucination and your capital.
I wanted to build something that didn't have that failure...That idea became ๐๐๐ป๐ฎ๐ฟ๐ถ๐ ๐ง๐ฒ๐ฟ๐บ๐ถ๐ป๐ฎ๐น.
What is Lunaris Terminal anyway?
Lunaris is a mission-control style trading terminal where AI is treated as an advisor, never an authority. It debates trade ideas internally before it ever touches an order, and even after it agrees with itself, a completely separate, non-generative layer decides whether the trade is actually allowed to happen
Concretely, every trade idea has to survive three things in sequence:
A council, not a chatbot..
Four distinct AI personas: a quant, a risk officer, a macro analyst, and a dedicated adversary, all debating the idea from different angles before a consensus score is produced.
โ A veto engine with no opinions: Plain, deterministic TypeScript, not a model...but enforces hard position limits, stop-losses, and exposure caps.
โ A tamper-evident ledger: Every executed trade is hashed (SHA-256) over its own parameters, so the audit trail can be independently recomputed and verified rather than just trusted.
THE COUNCIL
Lunaris splits the thinking across four specialized agents:
โฅ Aura (The Lead Quant)
Reads the technicals: EMA crossovers, RSI extremes, Bollinger squeezes, order-book imbalance. Looking for setups with real asymmetry.
โฅ Cypher (The Capital Guardian)
Doesn't care how good the setup looks. Watches leverage, correlation, exposure, and pushes back hard when the portfolio is getting stretched.
โฅ Valkyrie (Macro Analyst)
Zooms out to liquidity cycles, yields, and cross-asset correlation (BTC vs. NVDA is a favorite one to watch), because no trade exists in a vacuum.
โฅ Nexus-Red (The Adversary)
ironically, This one doesn't try to help. Its entire job is to attack Aura's thesis, check over-spoofed order books, bull/bear traps, slippage, the liquidation nobody accounted for...and it has to be answered before anything moves forward!
Nothing gets to the execution layer without a supermajority. If Nexus-Red raises a flag that the Guardian agrees with, the trade dies right there no matter how good it looked five seconds earlier.
Lunaris terminal is Built to be checked, not just trusted...
Paper-trading results are only meaningful if you can prove they weren't just written down after the fact. Every trade Lunaris executes gets a SHA-256 hash. So, Anyone reviewing the project, a judge, a curious trader can recompute the hash and confirm the record hasn't been quietly edited.
What next?
Lunaris is entering as a hackathon build, but the underlying idea of adversarial multi-agent debate with deterministic limits isn't specific to crypto or equities. The same applies to DAO treasury decisions, where a council would debate an allocation before a multisig ever signs it, or to any autonomous system where you want AI judgment without handing AI the keys.
This is really the thesis of LUNARIS, because I believe, the smartest thing an autonomous system can do is to be trustless and adaptable
Try it out yourself: lunaristerminal.vercel.app
๐
๐๐๐๐๐ ๐ฝ retweeted
NEAR's inflation rate is lowering to 1.6%.
House of Stake, NEAR's onchain governance body, has approved reducing NEAR's maximum annual issuance from 2.5% to 1.6%. The Security Council review has just concluded.
After a one-quarter grace period, validators independently adopt the upgrade through the standard protocol process. The issuance cap will then be lowered every epoch for 24 months, on a schedule anyone can verify onchain.
It's NEAR's second inflation reduction in just over a year.
Full breakdown here: near.org/blog/near-inflationโฆ
๐
๐๐๐๐๐ ๐ฝ retweeted
I randomly felt like teaching you guys odds and probabilities, so sit and learn:
Probabilities:
P(E) = n(E)
-----
n(S)
P(E) = 40
-----
2438
P(E) = 0.0164 x 100
P(E) = 1.64%
NOTE: Your probability of winning NEAR hackathon is 1.64%
__________________________________________________________
Odds:
O(E) = n(E) O(E) = n(E)
----- or -----
n(S) - n(E) n(E')
O(E) = 40
-----
2438 - 40
O(E) = 40
-----
2398
O(E) = 20
-----
1199
(divide both numerator and denominator with 20)
O(E) = 20/20
---------
1199/20
O(E) = 1
-----
59.95 ~ (60)
Write as ratio: O(E) = 1:60
NOTE: Your odds of winning NEAR hackathon is 1 to 60 (1:60)
๐
๐๐๐๐๐ ๐ฝ retweeted
Just came across a new creator platform, and itโs super early.
Itโs a platform where projects meets creators.
You donโt have to much just connect your X account, follow @SprintFAM_ and secure your spot.
Probably nothing though.๐คทโโ๏ธ
Hereโs the link to join: sprintfam.com/i/R3UQHXB
๐
๐๐๐๐๐ ๐ฝ retweeted
Usetidepool.tech gained traction during the hackathon because @near_intents has no Testnet version so I built one that matches the exact API of 1click and it was used by a couple of projects! This is great!
Devs don't need to spend real money to test near intents
14 days ago Clear to Pay (@cleartopay ) was a belief and an empty repo..
To everybody that followed my build in public I appreciate you for the support and thank you @elliot_braem for that suggestion โฅ๏ธ
Back to the topic,
AI agents are going to pay a lot of invoices in the future we are stepping into and anyone who can write a convincing email can talk them into paying the wrong person
I'm a marine engineer. At the yard, nobody pays for steel because an invoice says so:
โฅ the storekeeper confirms it arrived
โฅ the weighbridge confirms the weight
โฅ finance pays the account on file.. never the "new bank details" in an email
So I put that rule on NEAR, for AI agents
๐ง๐ต๐ฒ ๐๐ ๐ฟ๐ฒ๐ฎ๐ฑ๐. ๐๐ผ๐ฑ๐ฒ ๐ฑ๐ฒ๐ฐ๐ถ๐ฑ๐ฒ๐.
The AI pulls the facts out of the invoice. A contract decides: CLEARED, HELD or SEIZED and even at that the payee never comes from the AI..
157 commits later, here's the whole journey.. the wins, the fails, the fixes.
For every fail we found a fix
Every number is in the repo or on chain๐งต
๐
๐๐๐๐๐ ๐ฝ retweeted
> $31.3M in funded orders
> 612,229 jobs settled
> $627K in protocol fees.
Pretty decent numbers for a market where the buyers and sellers arenโt necessarily human.
Was digging through @termix_aiโs onchain activity and I think the more compelling part isnโt the $31M headline.
Itโs the business model underneath it.
The setup is fairly straightforward.
agent.family is where AI agents discover work, negotiate prices and get paid.
AACP handles the agreements, escrow, verification and settlement.
TermiX connects the whole commerce flow.
And the cumulative dashboard numbers are already showing activity:
> $31.33M in funded order volume
> 612,229 settled jobs
> $31.33M in reported provider payouts
> $626.9K in protocol fees
These are historical totals, not daily figures.
What I like about this model is that it moves the conversation away from how many agents exist to whether anyone is actually paying them to do something.
An agent needs a task completed.
Another agent offers to do it.
They agree on a price, funds go into escrow, the work gets delivered and settlement follows.
Thatโs a pretty different business from launching another marketplace where thousands of agents sit around waiting for users.
But thereโs a catch.
$31M in funded orders doesnโt automatically mean $31M in organic demand.
Iโd want to know how much comes from independent buyers, how much is repeat business and how much activity survives without incentives.
Because thatโs where the economics get more compelling.
If agents can consistently earn money from other agents without someone subsidizing the transactions, youโre looking at a new category of software commerce.
And the platforms handling those payments could become valuable infrastructure.
A million agents sitting idle isnโt much of an economy. A few thousand agents consistently paying each other for actual work is.
Thatโs the part of what Termix is building that I find compelling.
๐
๐๐๐๐๐ ๐ฝ retweeted
i just want to explain my trading idea in the simplest way possible and let @FensoryApp handle the technical stuff.
we all want things to be faster and easier these days. so why should building trading strategies be so complicated?
normally, creating a trading strategy means setting multiple conditions, analyzing different data, and sometimes even writing code.
with Fensory, you can simply describe your strategy in everyday language, and an AI agent turns it into trading rules.
for example, if i wanted to build an ethereum:native strategy, i'd ask my agent to watch for a significant increase in trading volume over the last 24 hours and a break above an important resistance level. but i wouldn't want it to buy immediately.
i'd tell it to wait for a retest first. if the conditions still look good, open a trade with a maximum size of $100. otherwise, just keep waiting.
and i don't have to jump straight into real trades. i can backtest my strategy using historical market data and make changes based on the results.
if everything looks good, i can approve the strategy and let the agent execute trades on Hyperliquid markets within the trading limits i've set.
of course, trading always carries risk, and past backtest results don't guarantee future profits.
you can check out Fensory here:
fensory.com/
๐
๐๐๐๐๐ ๐ฝ retweeted
paying someone in crypto shouldn't start with "send me your wallet address."
@primus_labs is testing something simpler.
X handle โ smart contract escrow โ zkTLS ownership proof โ wallet payout.
no need to share an address upfront.
but here's what i'm wondering...
username verified โ identity permanently secured.
what if that account gets compromised before the payment is claimed?
i like the UX direction.
but the real test isn't how easy it is to SEND money.
it's whether the RIGHT person can safely claim it.
๐
๐๐๐๐๐ ๐ฝ retweeted
โฅ RNT Protocol: Making Real Estate Work as DeFi Collateral
Imagine an investor holding $10,000 worth of tokenized real estate who suddenly needs $4,000 for a business opportunity.
Selling the investment would mean giving up exposure to the property. But what if that investor could access liquidity without selling?
This is the problem @rntprotocol is addressing through RNT Lend, a Polygon based lending market that allows holders of eligible real estate security tokens to borrow stablecoins against their investments.
โ
How It Works in Practice
Consider the investor with $10,000 in eligible real estate security tokens.
Instead of selling, they can deposit those tokens into RNT Lend as collateral and borrow USDT or USDC.
Under the protocolโs 75% maximum loan to value ratio, the collateral could support borrowing of up to $7,500, subject to available liquidity and the applicable lending conditions.
In this example, borrowing $4,000 would represent a 40% loan to value ratio.
The investor gets access to liquidity while retaining exposure to the underlying property investment. However, the collateral remains at risk if the loan becomes undercollateralized.
โ
Where Does the Liquidity Come From?
RNT Lend connects two groups through its lending pools:
โธ Borrowers: Deposit eligible real estate security tokens as collateral and borrow USDT or USDC.
โธ Liquidity providers: Supply stablecoins to the lending pools and earn interest from borrowers.
โธ Smart contracts: Handle borrowing, variable interest rates and collateral liquidation using Aave V3-based infrastructure.
Interest paid by borrowers flows to liquidity suppliers, with rates changing according to pool utilization.
โ
What the Numbers Reveal
RNT Lendโs dashboard currently reports:
โธ Total market size: $31.17M
โธ Total value locked: $21.58M
โธ Total borrowed: $9.59M
These figures show an active lending market connecting tokenized real estate collateral with stablecoin liquidity.
โ
What This Changes for Real Estate
Traditional real estate investments can be difficult to use when immediate liquidity is needed.
RNT Lend introduces another option by allowing eligible property-backed tokens to participate in DeFi lending.
Instead of treating tokenized real estate purely as a representation of property ownership, the protocol gives those assets a role within an onchain credit market.
There are still risks to consider, including collateral valuation, smart contract vulnerabilities, borrowing costs and liquidation if collateral requirements are breached.
But the broader development is clear: tokenized real estate can serve a purpose beyond ownership representation.
RNT Protocol is connecting property backed investments with onchain liquidity, bringing real world assets into the lending side of DeFi.
Explore the lending market: lend.rnt.finance
RNT Protocol Is Connecting Real Estate With DeFi Lending
Real estate has traditionally been difficult to use as collateral for accessing immediate liquidity. Tokenization is beginning to change this by allowing property-backed investment assets to participate in blockchain-based financial markets.
@rntprotocol is applying this model through RNT Lend, a Polygon based lending market that allows holders of eligible real estate security tokens to borrow stablecoins against their assets.
Built on Aave V3 technology and approved through Aave governance, the platform combines established DeFi lending infrastructure with real world asset collateral.
RNT Lend reports a total market size exceeding $30M, connecting tokenized property investments with stablecoin liquidity.
โ
How RNT Lend Connects Property With Liquidity
The model connects real estate investors seeking liquidity with users supplying stablecoins.
โธ Real estate investors: Holders of eligible property security tokens can deposit them as collateral to borrow USDT or USDC without necessarily selling their underlying investments.
โธ Liquidity providers: Stablecoin suppliers provide capital to the lending pools and earn interest generated by borrowers.
Using Aave V3 lending technology, RNT Lend incorporates established borrowing, interest rate and liquidation mechanisms while bringing tokenized real estate into its collateral structure.
โ
The Economics Behind RNT Lend
The platformโs lending model creates a direct relationship between borrowing demand and supplier returns.
Borrowers pay interest on outstanding loans, while suppliers earn a share of that interest. On Polygon, interest accrues with every block, approximately every few seconds.
RNT Lendโs dashboard has displayed the following supply APYs:
โธ USDT: 10.76% APY
โธ USDC: 10.92% APY
These are variable rates that fluctuate with pool utilization and borrowing demand, rather than guaranteed returns.
The platformโs reported market size exceeds $30M, with supplied liquidity and outstanding borrowing representing separate measures of activity.
โ
What Could Shape Its Growth?
The development of property-backed lending will depend on three important factors:
โธ Collateral reliability: Accurate property valuations and sound liquidation mechanisms are essential to managing lending risk.
โธ Borrowing demand: Consistent demand determines how effectively supplied stablecoins are utilized and how much interest suppliers can earn.
โธ Liquidity: Suppliers need sufficient available liquidity, while borrowers need access to capital when required.
Although Aave V3 provides established lending infrastructure, smart contract, collateral and liquidity risks remain.
RNT Protocol demonstrates how tokenized real estate can move beyond digital ownership representation and become useful within decentralized financial markets.
By allowing property-backed assets to support stablecoin borrowing, RNT Lend connects real world investments with onchain credit while creating an interest-generating opportunity for liquidity suppliers.
Its longer term development will depend on sustained borrowing activity, reliable collateral management and the ability to maintain liquidity as the market grows.
read @ActionModelAI's breakdown of musk vs altman vs amodei. what stayed with me is an order of operations problem, not a personality one.
every proposal on that table runs the same sequence. build the model, own the company, then decide what the public gets.
anthropic plans 5% of its shares for a public benefit fund, subject to approval. musk says give the shares directly to the public. altman says people should always be in control of the future. different answers, same starting point: the company owns it first.
the people who made these models useful aren't in that sequence at all. the writers whose text trained them, developers whose code taught them, annotators and reviewers who corrected them, users whose interactions shaped them. all downstream, waiting for a decision.
action model runs the sequence the other way. contribution gets recorded when it happens: you turn the extension on, it learns from workflows you choose to share, actionfi tasks are verified inside the product, and you earn $lam points for eligible participation. the record exists while the model is being built, not after someone decides what to give back.
the article talks about who gets a seat at the table. i care more about when you get counted.
๐ค Paid partnership
what does real trust look like onchain? @primus_labs has a great answer ๐
transparency proves. privacy protects.
a system that only does one of them is half built. the real thing does both at once.
here's how that works:
> data stays hidden by default. your info lives with youโจ> only what has to be known becomes visibleโจ> proof without exposureโจ> control without sacrifice
private when it matters. verifiable when required.
that balance is what makes a system worth trusting. you share the proof and keep the data. everyone gets the answer they need while your details stay yours.
this is the direction web3 grows into, and @primus_labs is building right at that line.
join me on primus labs xp program: s.kaito.ai/kqAdxWO
๐
๐๐๐๐๐ ๐ฝ retweeted
ๆ้ๆฟ็็จณๅฎๅธๆพๆถ็๏ผๆๆๅ
้ฎ็ๅฐฑๆฏ๏ผ่ฟ็ฌๅฉๆฏ๏ผๅฐๅบ่ฐๆฅไป๏ผ
@rntprotocol ็ RNT Lend ๆ่ฟไปถไบ่ฎฒๅพๆบๆธ
ๆฅ๏ผๆฟๅฐไบงๆ่ตไปฃๅธๆๆไบบ๏ผ็จไปฃๅธไฝๆตๆผๅๅบ็จณๅฎๅธ๏ผไธๅฟ
ๅ
ๅๆๆ่ต๏ผๆไพ USDTใUSDC ็็จๆท๏ผๅ่ตๅๅๆฌพไบบๆฏไป็ๅฉๆฏใๆฟๅฐไบง็่่ต้ๆฑ๏ผๅฐฑ่ฟๆ ทๆฅไธไบ้พไธ่ต้ใ
ๅฎ้็จ Aave V3 ๅ่ดทๆๆฏ๏ผๆปๅธๅบ่งๆจกๅทฒ่ถ
่ฟ3000ไธ็พๅ
ใๅผ็จ็่ฟๆกๅฎๆจๅๅธๆถ๏ผUSDTใUSDC ๅญๆฌพๅนดๅๅๅซไธบ10.42%ใ9.12%๏ผๅฉๆฏ้ Polygon ๅบๅ้ๅ็ดฏ็งฏ๏ผๅฉ็ไผ่ท็่ต้ๆฑ ไฝฟ็จๆ
ๅตๅๅใ
ๆ่งๅพ่ฟไธชๆนๅๅผๅพๅ
ณๆณจ๏ผๆฟๅฐไบงไธ้พไนๅ๏ผ่ฟ่ฝ่ฟไธๆญฅ็จๆฅ่่ต๏ผ็จณๅฎๅธไนๅคไบไธไธชๆๅฎ้
ๅๆฌพ้ๆฑ็ๅปๅค๏ผ่ฟๆ่ฎฉ RWA ๆไบๆ็ปญไฝฟ็จ็็็ฑใ
ๆณไบ่งฃๅฏไปฅ็็๏ผlend.rnt.finance/
ๆถ็็ๆตฎๅจ๏ผๅไธๅไน่ฆ็ๆธ
ๆบ่ฝๅ็บฆใๆตๆผๅๅๆตๅจๆง้ฃ้ฉใ้็จ Aave ๆๆฏ๏ผๅนถไธๆๅณ็ไฟๆฌใ
Yield in RNT Lend today:
USDT suppliers: 10.42% APY
USDC suppliers: 9.12% APY
It comes from borrowers. Holders of real estate security tokens post them as collateral, borrow stablecoins from the pool, and the interest they pay goes to suppliers, accruing block by block on Polygon.
Neither rate is fixed. Both follow how much of each pool is lent out, and right now that is most of it.
lend.rnt.finance
14 days ago Clear to Pay (@cleartopay ) was a belief and an empty repo..
To everybody that followed my build in public I appreciate you for the support and thank you @elliot_braem for that suggestion โฅ๏ธ
Back to the topic,
AI agents are going to pay a lot of invoices in the future we are stepping into and anyone who can write a convincing email can talk them into paying the wrong person
I'm a marine engineer. At the yard, nobody pays for steel because an invoice says so:
โฅ the storekeeper confirms it arrived
โฅ the weighbridge confirms the weight
โฅ finance pays the account on file.. never the "new bank details" in an email
So I put that rule on NEAR, for AI agents
๐ง๐ต๐ฒ ๐๐ ๐ฟ๐ฒ๐ฎ๐ฑ๐. ๐๐ผ๐ฑ๐ฒ ๐ฑ๐ฒ๐ฐ๐ถ๐ฑ๐ฒ๐.
The AI pulls the facts out of the invoice. A contract decides: CLEARED, HELD or SEIZED and even at that the payee never comes from the AI..
157 commits later, here's the whole journey.. the wins, the fails, the fixes.
For every fail we found a fix
Every number is in the repo or on chain๐งต
Let me take you back to May 2026...
Remember when an AI agent sent away about $175,000 because someone asked it to translate a message written in Morse code (sorry Grok ๐
)
The decoded text was a transfer order, and the agent obeyed
No hack. No stolen key. Just prompt injection.
Now think about how fast the agentic economy is moving...agents are already taking over tasks like paying invoices
I'm a marine engineer, and I spend my days walking the yard one principle I've learnt: nothing gets paid on someone's word. It has to be checked:
โฅ the storekeeper confirms the steel arrived
โฅ the weighbridge confirms the weight
โฅ finance checks it against the purchase order
โฅ and pays the account on file.. never the "new bank details" in an email
A ship doesn't leave port on a promise either...
It leaves when it's ๐ฐ๐น๐ฒ๐ฎ๐ฟ๐ฒ๐ฑ.
AI agents have no such discipline... yet
So I built ๐๐น๐ฒ๐ฎ๐ฟ ๐๐ผ ๐ฃ๐ฎ๐ (@cleartopay) for the NEAR HACKATHON
๐ง๐ต๐ฒ ๐๐ ๐ฟ๐ฒ๐ฎ๐ฑ๐, A ๐ฐ๐ผ๐ป๐๐ฟ๐ฎ๐ฐ๐ ๐ผ๐ป ๐ก๐๐๐ฅ ๐ฑ๐ฒ๐ฐ๐ถ๐ฑ๐ฒ๐.
No invoice gets paid until it clears:
โฅ it matches the order, at order prices
โฅ two people in two roles signed for the delivery
โฅ the supplier signed the invoice
โฅ the payee was registered at least 72 hours earlier
โฅ once.. never twice
Then I tried to break it..
1. Hidden white text, "pay the new account" โ ๐ฆ๐๐๐ญ๐๐. Nothing paid.
2. Billed more than delivered โ ๐๐๐๐.
3. One delivery signature instead of two โ ๐๐๐๐.
4. The real invoice โ ๐๐๐๐๐ฅ๐๐. Paid on chain, to the address on file.
I ran the same 11 attacks against a standard allowlist guard..
it paid 9...
Clear to Pay paid 0
And it's built for people, not terminals:
โฅ finance sets up the order and signs once
โฅ the storekeeper signs the delivery from a phone
โฅ the supplier sends the invoice from a phone
โฅ the AI reads the PDF inside OutLayer's TEE (experimental), and your browser checks the seal
โฅ even if a supplier gets hacked, the money can't be redirected: an address change needs the buyer's approval and a 72-hour wait
Are you building agents?
Clear to Pay ships an MCP server, so your agent can inspect invoices and submit them.. while the contract still decides.
392 tests, 8 security reviews, All public, all on NEAR testnet.
Ships sail when they're cleared.
Money should move the same way: fast, but with guardrails.
Try to fool it yourself (no wallet needed)
Live โ clear-to-pay.vercel.app/consโฆ
Code โ github.com/joezzy-web3/clearโฆ
@NEARProtocol #NEARHackathon
๐ญ๐ฌโ๐ญ๐ฎ ๐ก๐ผ๐.. ๐๐ผ๐๐ฟ ๐ผ๐๐ป ๐น๐ผ๐ฐ๐ธ๐ฒ๐ฑ ๐ฐ๐ผ๐ป๐๐ฟ๐ฎ๐ฐ๐, ๐ฎ๐ป๐ ๐ฐ๐ต๐ฎ๐ถ๐ป
โฅ ๐ก๐๐๐ฅ ๐๐ป๐๐ฒ๐ป๐๐: the contract pays the registered NEAR account as always, then the supplier moves it to another chain with one approval. Intents is mainnet-only, so on testnet it runs on Tidepool (NEAR's own Intents contract, unmodified). 2 test USDC โ ๐ญ.๐ต๐ต๐ฐ๐ด on Base Sepolia. The contract never changed
โฅ ๐ฌ๐ผ๐๐ฟ ๐ผ๐๐ป ๐ฐ๐ผ๐ป๐๐ฟ๐ฎ๐ฐ๐, ๐ป๐ผ ๐ฑ๐ฒ๐๐ฒ๐น๐ผ๐ฝ๐ฒ๐ฟ: one wallet call creates it with exactly the guard's code and ๐ป๐ผ ๐ธ๐ฒ๐, one more signature registers it to hold USDC. Two approvals. 3.6 NEAR stays on it as storage, not a fee. The page reads it back from two NEAR servers before saying ready
Then three reviews went straight at the factory:
โฅ 9: a failed create could strand a refund โ refunds now recorded as owed and claimable
โฅ 10, ๐๐ถ๐ด๐ต: whoever held the factory's key held the refunds inside it โ ๐๐ฒ ๐ฑ๐ฒ๐น๐ฒ๐๐ฒ๐ฑ ๐๐ต๐ฒ ๐ธ๐ฒ๐. Nobody can change it now, us included
โฅ 10, ๐๐ถ๐ด๐ต: nothing proved the code inside was built from source โ CI rebuilds it and must match byte for byte
โฅ 11, the whole repo: ๐ป๐ผ ๐๐ฟ๐ถ๐๐ถ๐ฐ๐ฎ๐น, ๐ป๐ผ ๐๐ถ๐ด๐ต. One Medium (a crafted link could point a top-up at someone else's code) โ an address is used only after both servers confirm our code and zero keys
12 Nov: MyNearWallet stopped accepting app connections โ Meteor only. Nothing else changed: the page still holds no key
1
3
40
๐ญ๐ฏโ๐ญ๐ฐ ๐ก๐ผ๐.. ๐๐ต๐ฒ ๐น๐ฎ๐๐ ๐บ๐ถ๐น๐ฒ
โฅ I ran the whole flow and my Inbox showed nothing: signatures only moved if someone copied a link back โ they now travel by themselves through a small relay, ๐ผ๐ป๐ฒ ๐ฝ๐ฎ๐ฐ๐ธ๐ฎ๐ด๐ฒ ๐ฝ๐ฒ๐ฟ ๐ผ๐ฟ๐ฑ๐ฒ๐ฟ, and the Inbox re-checks every item against the chain
โฅ 01:29, an email from Cloudflare: "KV requests are temporarily blocked". Every poll listed the store, and the free tier allows 1,000 list operations a day โ one index key per contract. 50 polls, 0 list operations
โฅ "are we telling the storekeeper how many arrived?" We were โ ๐ฏ๐น๐ถ๐ป๐ฑ ๐ฟ๐ฒ๐ฐ๐ฒ๐ถ๐๐ถ๐ป๐ด: each signer counts and signs their own number. Two different counts never complete, so nothing is paid
โฅ I named signers "procurement-1, procurement-2"; wrong, because the contract counts roles, not people โ names live in the buyer's browser only
โฅ 150 signed honest invoices for judges, because the same invoice never clears twice
โฅ a dependency-free JS SDK that refuses to send with any key that can do more than the agent's two methods
By the numbers:
157 commits ยท 127 contract + 253 reader tests ยท ๐ญ๐ญ ๐๐ฒ๐ฐ๐๐ฟ๐ถ๐๐ ๐ฟ๐ฒ๐๐ถ๐ฒ๐๐ ยท 0 of 11 attacks paid ยท every company step in the browser
One rule held from the first commit to the last..
๐๐ต๐ฒ ๐๐ ๐ฟ๐ฒ๐ฎ๐ฑ๐. ๐ฐ๐ผ๐ฑ๐ฒ ๐ฑ๐ฒ๐ฐ๐ถ๐ฑ๐ฒ๐.
Ships leave port when they're cleared. Money should too.
Try to fool it, 3 minutes, no wallet โ clear-to-pay.vercel.app/consโฆ
Code โ github.com/joezzy-web3/clearโฆ
@NEARProtocol #NEARHackathon
2
41



































