Verso

Been using gstack by @garrytan for a while now. The way it turns Claude Code into a full engineering team with slash commands is genuinely one of the best open source tools I've touched this year. So I built the marketing version. mstack: 21 marketing skills for Claude Code. Same architecture, same interactive workflow, same skill template system. Brand strategy, SEO, content writing, ad campaigns, social media - all unified by a persistent brand context so your AI never forgets your voice, audience, or positioning. And it collects zero of your data. github.com/VersoXBT/mstack

@adamdelphantom @colosseum well in case only bots in comments i'll help on prev hackathon I saw a project which focused on storing cum onchain after every goon which you had to tokenize. my personal mvp project ngl

What’s the weirdest project being built for @colosseum frontier?

@RoundtableSpace in 48 hours (5 months)

Someone gave Claude Code permanent memory and it hit 46k stars in 48hrs 95% less token consumption/session. Never hits context limits. Picks up exactly where you left off. One command install. Completely free.

@antoniayly @minara @lowesyang I can help with building that if you want, very experienced with different auto strategies for futures

@verso0x @minara @lowesyang I have to say yes minara strategy is a rather aggressive one. but it’s just hard to balance between PNL and levs and risks. thx for feedback we will keep improving that 🫡

The friend asked me to check @minara for trading At first glance all good, nice ui, profit strategies Then I took a look on positions (check screenshot) So you're telling me that for $100 profit you're going to risk $100k for 15 hours? I wonder what was max DD Same for other trades, you can find a bet that opens $300k position for $500 profit with holding time 15h. This is not a real strategy, don't get tricked by this, you'll lose your money. This needs MUCH improvements, absolutely unrealistic

@thedefiedge most of the volume sits on Polymarket and Kalshi Opinion is not even close

Prediction markets went from $15.8 billion in 2024 to $63.5 billion in 2025. Most of that volume sits on three platforms: Kalshi, Polymarket, and Opinion. Hyperliquid is about to enter the space. HIP-4 ships in weeks, and it puts perps and prediction markets in one account. The part that caught my attention is the capital efficiency angle. Your perp position and your prediction market position sit in the same collateral pool, so when one moves against you, the other offsets it. Same starting stack but w/ way more capital efficiency. The fee structure is worth looking at too: • Opening a position is free. You only pay when you close or settle (7 bps taker, 4 bps maker) • 1x isolated margin, no liquidation, max loss is what you paid in • Flip YES to NO in a single action. No slippage, no double fees • Burn a YES+NO pair to get your collateral back instantly To put that in perspective, Polymarket charges ~2% on winnings. Kalshi takes ~7%. Hyperliquid: 4 to 7 bps (0.04% to 0.07%). If you're actively trading prediction markets, that gap adds up fast. That said, market share doesn't flip on fees alone. Polymarket and Kalshi own 92% of prediction market volume between them. They're both well funded, and Polymarket has a loyal base of $POLY airdrop farmers waiting on the TGE. The thing I'm watching is whether the cross-margin angle is enough to pull active traders off the incumbents. Volume and market share over the next few months will tell the story. Shoutout to @Yaugourt who wrote the most comprehensive technical breakdown of HIP-4 I've seen. A lot of the details here come from his research and @androolloyd's work decoding the system from the hl-node binary.

@Shilllin whats up

sent dm yesterday but idk if you checked @mert so doubling here I have 2 versions for ts and rust: ts: gist.github.com/VersoXBT/073e1… rust: gist.github.com/VersoXBT/c7d8e… TS wins the average on my hardware, Rust wins on the busiest wallet. (I noted results in code as well, tested on a few different activity wallets)

announcing a mini solana dev weekend competition, $1,500 prize and bragging rights challenge: lowest latency algorithm for computing SOL PnL at runtime with no indexing and only RPC context: with existing solana RPC methods, you have getSignaturesForAddress and getTransaction the problem is gSFA only lets you traverse from recent -> old in one direction, so you are inherently capped on choice of algorithm with the new getTransactionsForAddress method however, you can start from start, end, middle, or any range you want and parallelize these calls to get 10x lower latencies the core challenge becomes "how do you search the set of transactions for an address the most efficiently given you do not know how sparse they are for a given wallet?" leave your submission as a DM or reply to this tweet as a gist or code sample and at the end of 2 days, I will run all the algorithms against each other for a set of addresses ranging from busy, sparse, and periodic. winner will be based on lowest avg latency

@binance @BinanceFutures Open-sourced the real-time validator + full Grafana dashboard here. Happy to share raw logs if you want to dig into the 13 restatements + 30%+ WS/REST mismatches.

Your backtest is lying to you. I ran live validation on Binance Futures OHLCV data for a few hours and found: - 13 candle restatements - 10 late-settling trades - WS/REST price disagreements on 30%+ of candles Built a framework that catches all of this in real time. 18 validators, live Grafana dashboard, fully open source. One-click docker setup. github.com/VersoXBT/live-…

Deployed personal website versox.dev so you know a little bit better who I am

@mert Tested a lot of variants DMed results @mert

@garryslist @garryslist digest for emails is broken, receiving empty

What if government worked for the people instead of a vocal activist class that virtue signals and grifts? That's the California we want to build. Join us.

@garrytan @garrytan digest for emails is broken, receiving empty

If you're OpenClaw or Hermes Agent-pilled and want to build your own Karpathy Knowledge LLM but are stymied by integrations or need to stop using grep on 10,000 files, I open sourced my personal Claw/Hermes schema, skillpack and RAG memory system github.com/garrytan/gbrain

@solana if you're currently wondering what to do with your life, the answer is almost certainly to build for the solana hackathon

if you're currently wondering what to do with your life, the answer is almost certainly to build for the solana hackathon

if you're currently wondering what to do with your life, the answer is almost certainly to build for the solana hackathon

context >>> everything

@garrytan Absolutely agree on that. Context architecture is the most important part of current models usage and makes big difference between users and ai engineers

Single russian participant I'm building in the Solana Frontier Hackathon, powered by @Colosseum! arena.colosseum.org/hackathon/soci…

@BitImmortal Ethereum is a stablecoin bro

since 2023 gold delivered higher returns than ethereum