Dinh Tran retweeted
The best AI design inspiration sites you need to save 🔖
- getdesign.ai/ - 𝙳𝙴𝚂𝙸𝙶𝙽.𝙼𝙳 files for AI agents
- posts.design - best of social post design
- loadmo.re - gallery of bold mobile websites
- recent.design - best of recent design, daily
- 60fps.design - UI animation & interaction details
- supahero.io - website hero section library
- navbar.gallery - the best navbar designs
- cta.gallery - call-to-actions that convert
- inspora.design - archive of recent visual design
Dinh Tran retweeted
强烈建议能力不差、还想多赚钱的人，都去做一次 Human 3.0。
这是油管大神 Dan Koe 提出的一套人生进化框架：心智、身体、精神、事业四个象限，每个都按 1.0 到 3.0 定位你现在的等级，再据此给你做未来规划，把整个人生往更完整的状态推。
有条件的话尽量用Claude去对话，效果会更好，让你的 AI按这套模型从四个象限给你打分，再用采访的方式一层层追问你的真实想法，最后给出一份只属于你的人生模型和进化指南。
你答得够具体、够真诚，一个问题一个问题耐心回，结果基本都会超预期。
Dinh Tran retweeted
you open X and see a startup you've never heard of. an hour later, a founder you follow is posting about it. then someone drops it into your team's Slack.
feels like magic from the outside, but it's an engineered 5-layer system with no ad spend:
1/ the founder anchor
raw product speed and narrative from the founder profile, never a faceless company account.
2/ pre-cleared creator drops
10 to 15 niche micro-creators locked in 48 hours early to post unscripted walkthroughs within the first 60 minutes.
3/ raw customer proof
unedited screenshots of speed, benchmarks, and active user reactions instead of polished marketing copy.
4/ the quote tweet network
3 pre-written angles given to founder friends so the post triggers immediate velocity in the algorithm.
5/ compressed attention waves
collapsing all distribution into a 4-hour window so the same 1,000 buyers see it 4 times in one morning. spread-out mentions get ignored; compressed waves create the illusion of inevitability.
steal the sequence and run it on your next drop.
Dinh Tran retweeted
兄弟们，教你们一个用 Claude 读钱包的野路子，我自己试了下，是真管用。
$780 滚到 $96,450，差不多 123 倍。不是靠押中了哪个神币，是照着一个牛逼钱包的打法抄作业。你换个地址一样能读，四秒就出结果。不用 API Key，也不用你会写代码，小白也能上手。
设置就一次，一分钟的事：
1️⃣ 打开 🔗claude.ai/ ，进设置，找到连接器，点添加自定义连接器
2️⃣ URL 那栏填 🔗mcp.bitquery.io/
3️⃣ 它会弹浏览器，你点个批准就完事了。不用你往哪贴 key，全程没密钥这回事，免费档就够折腾了。
这个连接器后面挂着 100 多个工具，覆盖 40 多条链。DEX 的交易基本秒级就索引进去了，几百万个地址标签，资金往哪流它都能给你追出来。
提示词直接抄，第一行把地址换成你想查的那个：
这里是一个钱包：<address>
拉取它过去30天内开过的每个仓位。
每个仓位都看：首次买入时代币多老、用了多大规模、拿了多久、是一次性卖还是分批出。
然后告诉我：它的赢家里有什么共同点，输家里有什么共同点。
不要预测，只看时机、规模、退出。
它会甩给你这些东西：
1️⃣ 真实的仓位规模，是有固定打法还是上头了瞎加
2️⃣ 真正进场的时候，那个币到底多新
3️⃣ 中位持有时间，不是他截图里想让你看到的那个数
4️⃣ 出货是一键砸盘，还是慢慢分批溜
5️⃣ 赢家都有、输家都没有的那点东西
看完这个再补一句，基本就扎到你自己心窝子了：
现在对我的钱包做同样的事：<your address>
两边放一起对比，告诉我最让我亏钱的三个习惯。
按亏掉的美元排，不按出现次数排。
地址上哪找，免费的：🔗fomo.family/r/NFTCPS
🔗gmgn.ai/r/C7KoyPop
给你看个 24 小时窗口的战绩：
1️⃣ +2,496,931.76 @The__Solstice
2️⃣ +974,887.26 @Natan_benish
我自己拿 342,482 个仓位全跑了一遍，说句大实话，答案从来就不是买哪个币。那条链上所有赚到的钱，87.8% 都是在代币出生头两个小时里落的袋；可就算这样，93.69% 的仓位到最后还是亏钱收场。
所以别老盯着 K 线图看了，多去扒钱包。图会骗你，钱包不会。
搞科研的、写论文的、做建模的注意了，有人做了个专门给 AI 用的画图 Skill，叫 Vivid Figures。
你不用管图叫啥名，把数据甩过去，说一句“比这几种方法，选能看清差异的图”，剩下的选图、画图、检查全是 AI 干。
我觉得最省心的两点：
1️⃣ 内置 108 个图表配方，山脊图、雨云图、热力图、三维曲面都能画，还能出 PNG、PDF 加源码，方便你自己改。
2️⃣ 五套配色随便换，鲜艳版稳重版两种风格，同一个任务里风格自动保持一致，不用一张张重来。
一句话：数据你给，脏活它干。
🔗 github.com/yjz211/vivid-figu…
Dinh Tran retweeted
this is where the AI shift gets f*cking serious
297 pages on building a $100K/month business with zero employees. 8 parts, 40 chapters, 50 links
most founders still pay people to research leads, update CRMs and write reports while the AI sits on the sidelines making suggestions
the stack it walks you through:
> Kimi K3 - the brain behind the business
> Kimi Code - the hands that do the work
> Skills - encode your best employee once
> MCP - keys to the systems that matter
> AgentSwarm - scale to a full workforce
growth used to scale with payroll
now it scales with how many expensive workflows you can rip out without breaking anything
one founder + the right systems = $100K/month without the payroll that used to come with it
Dinh Tran retweeted
this is absolutely unreal
jack dorsey (twitter co-founder) just shipped a totally free github repo sitting at 26.2k stars that basically hands you an ai-agent operating system for running a business
here’s the setup:
1. clone the repo
2. host the server yourself: channels, search, git, automations, it all runs there
3. drop your agent into a channel like it’s another teammate, lock down its permissions, and let the crew guide it in real time
save this and bookmark it, seriously
Dinh Tran retweeted
this is f*cking gold
How to build your first AI agent (Full guide)
if I had this a year ago, I would've shipped my first agent in a day instead of 2 weeks in the right hands, this changes everything:
Dinh Tran retweeted
All Web3 cards in one place.
There are more and more cards, we have collected everything for you in one place so as not to lose.
@ether_fi:
• Core is free (but requires KYC). FX 1% (0% USD/EUR), 0.2 deposit%;
• Non-custodial Safe + loan mode, up to 3% cashback (for Free limits);
• Campaigns: food/taxi bonuses, upgrade to metal cards.
@useTria:
• Virtual is now free ($25/year, but now fee waived). 0% deposit, FX ~1%;
• Self-custodial MPC, 1000+ tokens, up to 6% cashback + APY;
• Campaigns: virtual free, July Spend Fest (doubled limits).
@gnosispay:
• Free version. 0% for transactions/FX, ATM for free up to €200, then 2%;
• Fully self-custodial Safe, ENS-name on the card, up to 5% in GNO.
@RedotPay:
• Base card is $10.0% per month. 0-1% crypto replenishment, spending ~1% + FX 1.2%;
• Very high limits ($1 million/day), German IBAN, XRP loan;
• Promo: 20% discount on the card, Pro-subscription with cashback.
@Tangem:
• You need to buy a card (starting from $50 per set) / tariff: $0 (but there are country restrictions) / Transaction fee: $0
• Seedless hardware wallet + Tangem Pay (self-custodial USDC Visa).
• Campaigns: welcome-bonuses up to $75 (coming soon).
@KASTxyz:
• Standard for free. Replenishment with stables 0%, USD 0%, FX 0.5–1.75%;
• 1.5–3% cashback in USD according to the tariffs;
• Campaigns: summer 2× cashback, 10% on advertising, 0% FX, 10% on PUBG.
@Plasma:
• Lite for free (virtual). Replenishment with 0% stables, FX ~2%.
• Self-custodial on your own L1, XPL 2-4% cashback + Earn Vault.
• Campaigns: $10 for the first $100 spend, increased AI cashback.
@Exa_App:
• Free. FX 1,5%, Fee 0%;
• On-chain credit card: installments of up to 6-9 payments without asset sales + profitability.
@Bybit_Official:
• Virtual for free. Registration of $0 / Year / month $0 / Crypt conversion 0.9%. FX 0,5–2%;
• Exchange-traded Mastercard, 2-10% cashback on VIP + 100% rebates on Netflix/Spotify.
@wirexapp:
• Basic is free. FX 0%;
• Up to 50% cashback on popular services such as Uber One, ChatGPT, Claude, etc.
@raincards:
• Often 0% per annum (depends on the partner), FX up to 3%, 0-1% conversion.
• Infrastructure for white-label stablecoin cards (Avalanche, etc.).
@xplaceapp:
• Basic is free. FX 1%, cashback 0.5% ($2000 limit on the Free plan).
• Non-custodial loan for Solana (Kamino): collateral → USDC credit line + yield.
@Krak:
• Free (virtual & physical). 0% transaction / FX
• Campaigns: Welcome Max tier (2% cashback) for first 30 days
@Nexo:
• Free (virtual with $50 balance; physical free for Gold+). FX 0.2% (EEA/UK/CH weekdays) / 0.7% weekends
• Dual-mode Mastercard (Debit: spend balances while earning up to ~13–14% APY
• Campaigns: Platinum tier on arrival (30 days)
@coca_card:
• Starter is free. FX 0%, ATM is free up to $250, then 2%.
• Non-custodial MPC, up to 8% USDT + 6% APY (Morpho), EUR IBAN.
@avici:
• 4 free virtual, then $10. Transactions 0%, FX 0% (Visa ~0.4–1%), ATM $0-1+0.65%;
• Self-custodial secured credit, passkey, SOL-backed line of credit with profitability.
@hyperbeat:
• Maps are free. Deposit ~0.12%, cashback up to 5% (campaign), portfolio loan;
• Liquid Banking on Hyperliquid: credit + profitability + trading. Campaigns: up to 5% cashback.
@BleapApp:
• Completely free. FX 0%;
• Non-custodial Mastercard (account abstraction), categorical cashback up to 20% USDC (streaming/AI).
@oobit:
• Free of charge. Transactions of 1% (min. $0.25), FX up to 3%.
• Payment directly from an external wallet, up to 10% cashback, Agent Cards for AI.
@slashapp:
• Free plan. 1% FX cards (min. $0.4), transfers of $1-25;
• Business: unlimited virtual cards + Global USD (without an American company).
@FlexSuperApp
• Creation is free (next $3). FX $0.30+1%, replenishment is free (except Tron);
• Business: banking + cards + cost management + stablecoins.
@perena:
• Withdrawal of 0.05%, card fees;
• Yield-bearing USD* (delta-neutral) on Solana + spending in the ecosystem.
@altitude:
• Virtual business for free (unlimited). 1-2% USDC Cashback;
• Unlimited business cards, programmable treasury.
@PulsarMoneyApp:
• Early access. 0% deposit, same currency 0%, cross currency 1.35%;
• USD/EUR accounts + Virtual card + ~5% APY + Social payments.
@DptPay:
• Virtual $10 / Physical $50. USD 0%, FX 0–2%;
• Stablecoin Visa + DeFi-profitability on balance, travel insurance up to $1.5 million.
There are a lot of cards, but in our experience, EtherFi, KAST, Bybit + Tangem are most often recommended.
Dinh Tran retweeted
高频做市基建二下单和风控
聊完服务器层面，说下单。
下单层踩过最大的坑是以为下单就是发请求。其实不是的。下单是一个完整的链路，从策略决定报价到交易所确认成交，中间每一环都有延迟，每一环都可能出事。
我们现在的架构底线：
第一，核心报价走 WS，不用 REST。REST 是给你查余额用的，不是给你抢单的。哪怕是批量下单，也必须走 WS。利用本地 buffer 把这一次需要下单和撤单的指令打包存好，然后一次性 flush 出去，速度比走 REST 快得多。
第二，API 限流（Rate Limit）的硬隔离。不做市可能不知道，交易所的限流分两种情况：锁 API Key 和锁 IP。这两者的解决方案完全不一样。如果交易所是锁 API Key，用基础的令牌桶（Token Bucket）就可以了，给每个服务分配额度，防止一个策略抽风把 token 跑满导致全家卡死。但如果是锁 IP，光靠基础令牌桶就不管用了，你必须在底层建一个 IP 桶（IP Pool），一旦触及阈值立刻做动态切换。
第三，执行 p50 控制在 30us 以内。什么概念？从策略发出指令到网卡把包推出去，中间不超过30微秒。超过这个数，你的单子在盘面上就是慢的。
第四，网络分离。策略的发单绑定 EIP 发送订单组，这跟接收行情的 ENI 必须分开。不然遇到极端大行情，海量数据瞬间把带宽打满，你的撤单指令根本挤不出去，只能看着单子被吃。
再说风控。
风控和策略一定要分开。
我们的做法：行情延迟超过阈值（比如 500 毫秒），直接熔断不下单，并且立刻撤单。
为什么？不仅是因为“信息过期”500ms 的延迟意味着交易所服务器本身丢包或者通道严重卡顿。这个时候，交易所内部大概率已经发生了撮合，只是数据包还没推给你。如果你继续按手里的残影去强行定价，算出来的价格必然是错的，挂出去就是坏价。所以这种时候不仅不能挂单，还要立刻撤单防守。这不是策略问题，是底线风控问题。
Kill switch 单独管理最大回撤，和策略层完全解耦。策略层管赚钱，风控管别亏死。两个不能混在一起写，因为策略出 bug 的时候，你得能一键拔掉，而不是去策略代码里找开关。
Dinh Tran retweeted
every App Store rejection costs you 2-5 days.
so we built a cli that scans your iOS app against apple's guidelines before you submit.
> payment & IAP compliance
> privacy manifests & data declarations
> required sign-in & account deletion flows
> metadata & completeness checks
> binary validation
now with a added cloud device feature, so it validates full user flows, not just the binary.
made it a claude code and codex skill. it fixes every issue it finds. scan, fix, repeat until it passes.
Dinh Tran retweeted
A 20-year-old in Shenzhen owns a company that turned $1,200 into $395,000.
Its workforce is 300 AI agents. Its CEO is also an AI.
Human employees: zero.
The entire org chart:
> CEO - Claude Opus 4.8. Sets the goal, vetoes every weak trade.
> Workforce - 300 agents. They read the order book, the news, every Bitcoin market at once.
> Risk officer - a hard daily loss cap. Shuts the whole floor the second it's hit.
> Payroll: $0. Meetings: 0. Days off: 0.
> HQ: a dorm room.
It runs as one loop - the agents propose, the CEO disposes, the survivors trade, repeat. He just owns the place.
Save this full loop pipeline is in the post below.
🤖 Made with AI
While AI timeline argues about loop engineering, a 20-year-old student in Shenzhen built the biggest loop anyone's aimed at money: 300 agents, one brain, $1,200 -> 395,000.
Not a prompt. Not a bot. A loop that runs the whole job on its own - and never asks him a thing.
No fund. No backers. Just a laptop, a dorm room, and a bet that a swarm beats a genius.
His wallet: polymarket.com/@easypredict?…
The smartest single model still works one step at a time. His loop moves in 300 directions at once - and never checks in with him.
Claude Opus 4.8 runs the floor as the brain. 300 agents are the hands. One loop ties them together:
> Decompose - Opus breaks the market into targets
> Dispatch - 300 agents scatter across them at once
> Execute - thousands of reads in a single pass, not one by one
> Review - Opus grades every edge before a cent moves
The hands find. The brain judges. The loop learns what failed and runs again, tighter each time. One daily loss cap, and it shuts itself down.
It's not a bot you babysit. It's a loop that runs the whole desk.
Everyone's just learning to run one loop. He runs 300 in one.
The edge was never the genius model - it's the size of the loop, and the brain in charge of it.
Save this - the full 300-agent loop is laid out above, start to finish. You'll want it when you build your own.
300 agents are looping right now, hunting the markets everyone skips. Two clicks and their trades land in your wallet too: kreo.app/@cvxv666
Dinh Tran retweeted
How a 24-year-old programmer from Portugal made $18,200 in a month on football betting
He created an AI analyst that finds flaws in bookmakers' live lines in real time and delivers predictions with an 84% win rate.
Costs: $0 (Used free APIs and Windsurf IDE)
He launched a Python script that maps out match videos in real time:
Top layer: A Computer Vision algorithm recognizes the positions of players from both teams (blue and pink dots) and the ball, instantly transferring them onto a 2D pitch layout. This allows the AI to track team formations and open spaces in high detail, things regular bettors completely miss.
Bottom layer: Python code (written alongside the Windsurf AI assistant), where the SoccerPitchConfiguration class defines the field, penalty box, and center circle dimensions down to the centimeter for perfect player-distance calculations.
The AI constantly correlates the real-time movement of players on the pitch with live bookmaker odds. The moment the algorithm detects that a team has pinned their opponent into a specific zone or exposed their flanks, while the bookmaker hasn't adjusted the odds yet, the script automatically fires a betting signal.
First week:
>Live bets placed: 142
> Won bets: 119
> Net profit: +$4,350 using a flat $50 stake
The AI completely automated the entire cycle: Windsurf and Claude wrote the tracking code, and the algorithm autonomously parses live odds, calculates the mathematical expectation of value bets, generates player heatmaps, and spots hidden tactical anomalies. It runs 100% autonomously.
Bookmark it and check out the article below 👇
Dinh Tran retweeted
FOUNDERS: You've been paying to build on Claude. @AnthropicAI launched a program to change that.
@Claudeai for Startups - free API credits and priority rate limits for early-stage VC-backed founders:
- Free Claude API credits
- Highest rate limits, no throttling in production
- Hackathons, Founder Days, and meetups
- Early access to new model releases
Build with the full Claude stack:
Claude API, Claude Code, Claude Managed Agents, and Claude Cowork.
To qualify: your startup must be early-stage and backed by one of Anthropic's partner VCs. Ask your investors for a unique application link.
Apply → claude.com/programs/startups
P.S. Founders using Claude to build - when you're ready to raise, @ThePageform is where your data room lives → pageform.io
Dinh Tran retweeted
Web scraping is dead.
PixelRAG skips HTML parsing completely. It screenshots the page and a vision model reads the answer straight off the pixels.
100% open-source. Comes with a Claude Code plugin that gives Claude eyes.
Dinh Tran retweeted
THIS SINGLE CLAUDE.md FILE HIT #1 ON GITHUB TRENDING.
90k+ stars. 7.8k forks. zero dependencies.
save this before the feed buries it.
it’s not another prompt pack.
it’s a simple CLAUDE.md file built around 4 principles inspired by Karpathy’s coding habits:
/ think before coding
/ keep it simple
/ make surgical edits only
/ define the target before you start
drop it into your project today and Claude Code starts behaving very differently.
read it today. link below.
Dinh Tran retweeted
I genuinely don't understand why everyone isn't using this yet
Andrej Karpathy, a co-founder of OpenAI, posted a simple idea that hit 16 million views: stop using AI to write code, use it to build a second brain.
You point Claude Code at a folder, drop in any source, an article, a transcript, a PDF, and Claude reads it, links it, and files it into a living wiki of everything you know. It compounds like interest, the more you feed it, the smarter it gets.
Here's the whole thing:
> Install Obsidian, create a vault, open it in Claude Code
> Paste Karpathy's wiki idea file and tell Claude to build it
> Claude makes three folders: raw for sources, wiki for its pages, a CLAUDE.md that runs it
> Drop any source into raw and say "ingest this"
> Ask questions across everything, forever
Five minutes to set up, and you never start from a blank chat again.
Full step-by-step guide with Claude and Obsidian, link below.
Bookmark this
