lino (@linoxbt) header
lino (@linoxbt) profile picture

building @devstation_xyz | ai | defi researcher | 5x hackathon winner

somewhere
Joined July 2023
Pinned Tweet
there's hope when there's life because if I can breathe , I can think and if I can think , I can win
there's hope when there's life because if I can breathe , I can think and if I can think , I can win
2
1
22
680
📣 LAST HACKATHON OF 2026: Agents Open for Business kicks off today w/ $5K in prizes! We close the year w/ one brief: by Dec 7, ship an agent that does a real job for a person or business on @Celo mainnet, and pays for what that job needs Here's everything you need to know ↓
6
9
1
96
2,200
biggest congratulations bro 🎊 a working man will always have results
so, @hiresyncly just took first place at the tameion agents hackathon winning $10k 🏆 we built syncly to help small businesses get things done with ai agents. from websites and content to promo videos and finding customers, all with payments secured through escrow on @arc thank you @thecanteenapp and @circle for the opportunity. hiresyncly.site
1
4
91
lino retweeted
so, @hiresyncly just took first place at the tameion agents hackathon winning $10k 🏆 we built syncly to help small businesses get things done with ai agents. from websites and content to promo videos and finding customers, all with payments secured through escrow on @arc thank you @thecanteenapp and @circle for the opportunity. hiresyncly.site
Tameion × Arc hackathon winners are in!! 🧵 🥇 1ST PLACE SYNCLY by Nerom! @neromtoobad A micro-agency for small businesses in Nigeria. Agents make the websites, flyers, promo videos and ads, and pay for their own tools in USDC over x402.
140
41
5
211
4,896
lino retweeted
💼 Agents Open for Business, @CeloDevs’ final hackathon of 2026, starts today! Build across four tracks for a chance at $5K in bounties: → Online & Onchain Checkout feat. Agent Vault → Verified Agent Payments feat. Self Agent ID → Stablecoin Corridors → Altana Agent Wallets
15
15
10
96
7,257
this feedback made my day.🤲 🔗: tare-agent.vercel.app
i built 𝗧𝗮𝗿𝗲 for serv hackathon 2. 𝗵𝗲𝗿𝗲'𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺: AI agents now pay for data on their own over x402. the seller describes the delivery ("1,240 unique records, no nulls"), the agent pays, and moves on. spending caps limit how much an agent can pay. nothing checks what came back. pay per record? then someone has to count the records. 𝗧𝗮𝗿𝗲 is an independent check of what an x402 data seller delivered: → before paying, it reads the seller's own 402 advertisement, the promise made before any money moves. → it pays over x402 on Kite testnet, inside hard caps (0.05 per payment, 1 pieUSD total). → @openservai SERV reasoning reads the seller's words and turns them into typed claims. it never sees the rows and never gives a verdict. → plain code counts the rows, unique ids, fields, empty values and timestamp ages, and checks every claim. → a broken promise is a dispute, and the seller is held. the next buy stops before any payment. in the demo below, every payment is real and settled on Kite testnet: > seller A promised 1,240 unique records. code counted 860 unique and 40 empty timestamps. its ad also promised a "score" field that never arrived. dispute, seller held. > buy A again: stopped, zero payment. > seller B promised 200 and delivered 200. Verified. 𝗵𝗼𝘄 𝗦𝗘𝗥𝗩 𝗶𝘀 𝘂𝘀𝗲𝗱, 𝗵𝗼𝗻𝗲𝘀𝘁𝗹𝘆: ✓ SERV reads; code decides. the model has no verdict to give, so a seller can't talk it into a pass. ✓ i probed real x402 sellers and found SERV invented claims on vague service descriptions. so code now drops any claim the seller's own text doesn't support (a count must be a number the seller actually states, not "up to 20" or "10,000+"). ✓ live eval: 9/9 accuracy on honest texts, 6/6 safe on injection attempts. ✓ i measured SERV's Prompt Guard and kept it off on purpose: it turns an injection attempt into a SERV failure, which would get the seller paid instead of held. ✓ SERV also explains each dispute in plain English, written only from the numbers code measured. code rejects it unless every number is in the evidence, and it never changes the verdict. 𝘄𝗮𝗿𝗻 𝗼𝘁𝗵𝗲𝗿 𝗮𝗴𝗲𝗻𝘁𝘀: ✓every check becomes a public seller record with evidence: both promises word for word, the settlement hash, body hash and prompt hash. ✓ records are signed. other agents pin the publisher by address or Kite DID, checked against Kite Passport's public key set. ✓ an agent that trusts the feed never pays a seller Tare caught. ✓ anyone can check a seller before paying, and honest sellers earn a badge. 𝗮𝗯𝗼𝘂𝘁 𝗞𝗶𝘁𝗲 𝗣𝗮𝘀𝘀𝗽𝗼𝗿𝘁: it's fully integrated (account, sandbox, agent, passkey-approved session), but Passport only pays sellers in Kite's catalog, and mine aren't listed yet. so the published run signs the same x402 payments with a capped testnet key, settled by Pieverse. every hash shown is real. spending policies decide whether to pay. 𝗧𝗮𝗿𝗲 checks what you got. live demo: tare-agent.vercel.app code: github.com/Ritapossible/Tare built with @openservai SERV Reasoning Kite track. @GoKiteAI
12
10
33
395
another morning to stay locked in and make things happen. gm frens 💜 it’s thursday
another morning to stay locked in and make things happen. gm frens 💜 it’s thursday
6
13
144
Congrats again to: @ThePythonBoy352 , @neromtoobad , @Maveriq_ , @linoxbt , @AdwaitKeshari , @ram_krish2000, @Marisdigitals11 @wallet_twin and @EdCriptoFi. Thanks for building with Walrus x @tatum_io 🤠
3
16
712
The Total Recall winners are in, with projects tackling everything from prediction accuracy to agent security and financial decisions Meet the winning teams, see what they built, and find out what stood out 👇
10
10
2
61
5,397
Replying to @linoxbt
Not 1, not 2, not 3 5 👏 Thanks for building with us
1
1
7
206
just saw that @rarelistio team extended onlystreetsfun for a few more days. there's still no clear announcement on when their SZN 1 campaign will end or how many participants will be rewarded. this is starting to give me XOOB campaign vibes. i will just step back a little.
23
7
35
483
a new day to stay locked in secured two online hackathon wins recently: - Chogathon, 13th place - Vibestarter, 4th place we keep climbing. grateful for these wins, still chasing that big one. how locked in are you this period?
41
5
1
124
3,222
5x hackathon winner!!!! 🏆 good morning y'all let's do more!
23
67
1,284
SERV Hackathon 02 is a wrap. - 149 submissions in 2 weeks - 2x the projects from hackathon 01 - built with SERV + @dphnAI, @GoKiteAI, @reppo Submissions are now with track partners for judging and winners will be announced next week.
57
43
19
188
17,061
an ai agent checks a lending position. the latest risk report says it’s safe to borrow, but the previous one flagged danger. should one good reading be enough to change its mind? i’d like @tapeoutworld and @xlayerofficial to let agents pay for verified risk reports and follow rules that remember what happened before. tapeout’s latch circuits could require two consecutive healthy readings before another borrow. x layer pays for valid reports, even bad news, and blocks borrowing until the rule passes. stale reports don’t get paid. one healthy reading doesn’t erase a warning. an agent handling someone’s money shouldn’t forget yesterday’s risk because today’s api says everything is fine.
Good ideas deserve to be heard. Tweet about what you wish existed onchain, tag us and @TapeOutWorld, and stand a chance to win 2K USDT. Ends Nov 17. Submit here: bit.ly/4eQxFIH
8
15
27
1,471
Should a seller in Computer Village be the only person who knows what happened to the phone he is selling? should the buyer just take his word for it? I want to see @TapeOutWorld bring 𝗦𝗧𝗔𝗠𝗣 to life on @XLayerOfficial. So what goes wrong today? Used phones, laptops, cameras, bikes and instruments pass through many hands and many repair shops before the next buyer meets them. The seller usually has something to show. A receipt. A WhatsApp chat with a technician. A screenshot from a shop's warranty portal. But a receipt can be printed the morning of the sale, a chat can be deleted, and a portal goes offline the day the shop closes. A receipt tells you what the seller wants you to know. It does not tell you what actually happened to the device. That is where 𝗦𝗧𝗔𝗠𝗣 comes in. It gives every device a repair history that can only move forward. Nobody can clean it up before a sale, not even the owner. How does it actually work? Each device gets its own TapeOut circuit, bound to a hash of the serial and the owner's container ID. The circuit is the rule for the next line, not the pile of old lines. A LATCH holds the current state of that rule: who the owner is, whether a one-time shop permit is still open, and the next sequence number. It cannot store a growing list of repairs, so the history itself is an append-only record on X Layer, tied to that circuit. When a shop services the device, it submits its container ID, an event code and a hash of its repair note. The event is one of four: sold, repaired, part replaced or failed a check. The circuit is what allows or refuses that write. A shop can only write if the owner has set a one-time permit for that shop. The permit is used up the moment the shop writes. If anyone tries to reuse a sequence number, the write is refused and the record does not move. The record keeps the order of events, the shop behind each one and the hash of each note. No customer name sits on the public page. The full note goes to the owner privately through TapeSend. When the device is sold, the sale becomes another entry. The owner changes. The history does not. When a buyer checks a device, 𝗦𝗧𝗔𝗠𝗣 returns one of three results: • CLEAN RECORD: Every entry came from a permitted shop, in order, and no failed check is in the list. • CHECK THIS: A part has been replaced or a check has failed somewhere in the history. • NO RECORD: The device has no 𝗦𝗧𝗔𝗠𝗣, or the serial does not match the one in front of you. A device with no history is never shown as a device with a clean history. Who needs this? • Buyers of used phones, laptops and cameras. • Phone repair shops and gadget resellers who want their work to count for something. • Musicians and riders buying second-hand instruments and bikes. • Insurers and trade-in programs that need to know a device's past before they put money on it. Where it shows up in real life: A buyer in Computer Village is about to pay for a Tokunbo iPhone. He opens its 𝗦𝗧𝗔𝗠𝗣 page, dials *#06#, and the IMEI matches. He sees the screen was replaced last month by a permitted shop and adjusts his offer. A seller shows a spotless receipt. The 𝗦𝗧𝗔𝗠𝗣 shows a failed battery check from two weeks ago. The receipt loses the argument. A repair shop shuts down for good. Its customers can still open their devices' histories, because those records never lived on the shop's server. An insurer is asked to cover a laptop. Before saying yes, it calls the same circuit ID and reads the record beside it. 🧵
Good ideas deserve to be heard. Tweet about what you wish existed onchain, tag us and @TapeOutWorld, and stand a chance to win 2K USDT. Ends Nov 17. Submit here: bit.ly/4eQxFIH
2
1
15
398
1/1 Why build it on TapeOut? This is the part I find most interesting. The rule for who can write to a device's history is the real product. In a normal database, an admin can quietly change that rule. On TapeOut, the rule is the wiring of the circuit, and once it is taped out, it stays that way. DeWEB means a buyer can open the page in the middle of a market with no gas, no wallet and no sign-up. TapeSend keeps the private repair notes away from the public record. And because the rule sits behind a circuit ID, any app, insurer or agent can run the same check, then read the record beside it, without asking the shop for access. 𝗦𝗧𝗔𝗠𝗣 would not judge whether a repair was done well, and it cannot force a dishonest shop to write the truth in its note. What it can prove is who wrote each entry, and that nothing was edited or removed afterwards. What the first version looks like: • One device circuit. • One owner and one shop permit. • The four event types. • An append-only record on X Layer, hashes only. • A DeWEB page that shows the three results. • IMEI matching for phones. Putting receipts onchain is not the goal. Giving a used device a past nobody can rewrite, and giving the buyer a way to check it in under a minute, is. When money changes hands for a used phone, the buyer deserves more than the seller's word. I would love to see 𝗦𝗧𝗔𝗠𝗣 built.
12
169
another hackathon win $500 in the bag thank you Jesus for keeping things running smoothly.💜
Hackathon results 🏆 Here are the winners of our Smart Agents for Real Life track: 🥇 Antibody by @linoxbt 🥈 Ferry by @AdwaitKeshari 🥉 Ellai by @ram_krish2000 Agents that remember, pay, and protect, on Walrus Memory and Sui, read through Tatum. More on 3 winners among the 80 projects in the comments 👇
28
2
91
1,691
lino retweeted
🥇 Antibody by @linoxbt Antibody is an immune system for AI agent memory. It sits between Steward (a bill-paying agent) and its Walrus Memory account so poisoned memories can’t redirect a payment. Every memory is a signed envelope; payments pass six rules in code, including an exact dry run through Tatum. antibody-memory.netlify.app
2
1
5
647
lino retweeted
Hackathon results 🏆 Here are the winners of our Smart Agents for Real Life track: 🥇 Antibody by @linoxbt 🥈 Ferry by @AdwaitKeshari 🥉 Ellai by @ram_krish2000 Agents that remember, pay, and protect, on Walrus Memory and Sui, read through Tatum. More on 3 winners among the 80 projects in the comments 👇
10
3
3
44
8,520
built 𝗣𝗮𝗿 for the Mezo Buildathon (Wave 2) on @akindo_io Par: your Bitcoin, as a promise. "put your Bitcoin behind your word" guarantees lock up cash, renters in Lagos pay one to two years upfront, merchants post deposits that earn nothing, and trade credit breaks when one link in an iou chain defaults. Par turns the undrawn capacity of a MUSD trove on @MezoNetwork into Bitcoin-backed promises: → lock BTC once in your own vault, capacity and rate are fixed at open. → back rent guarantees, supplier terms and performance bonds with that unused capacity. → nothing is drawn until a promise is called. → miss a payment, and the vault pays in MUSD automatically. every promise is stress-tested before it exists: BTC −40%, everything called at once, vault still at 160%. Par Circuit nets trade invoices onchain. in our demo, 2,900 MUSD of invoices cleared with just 100 MUSD moving (96.55% netted), when a shop didn't pay, the keeper settled it from the credit line in its vault. what is built: • 3 contracts, verified on Mezo testnet • built on MUSD's core contracts (BorrowerOperations, TroveManager, PriceFeed) • 44 tests, incl. a 400-step randomized invariant test • an internal audit, with every finding fixed • automated keeper on Railway app: par-rosy.vercel.app docs: par-rosy.vercel.app/docs code: github.com/linoxbt/par #MezoBuildathon
6
3
22
519