Where your Solana / Rust bug gets caught: trusted types → in your editor const asserts/lints → when you compile pinning tests → when you run tests runtime checks → in production ... hopefully Move everything you can up the list. Read how at

Our co-founder and CEO @Josh_E_Wa shares what 4+ years at @dydxfoundation taught him and how those lessons are shaping our upcoming perp DEX on @solana: • Be deliberate about what we build ourselves • Earn every trade • Build alignment early • Keep the team small and ownership real • Raise what we need. Build toward break-even

On-chain matters because it enables: - Verifiability: you actually know the rules of the venue you’re trading on vs ”trust mes”; can build venues without unilateral oracle price changes or first look by an internal maker. You also can get some execution guarantees. - Atomic transaction composability: ex. can open a position on separate venues in one all-or-nothing transaction, can move funds between protocols without brittle bridging. - Lower risk of censorship/exploitation: you can build, choose, find protocols that have frozen admins or more distributed admins; would need tens of billions of capital to properly capture chain stake and censor; greatly lower venue risk. - Long term alignment: stable API with real money behind it, many stable users even as top app changes, many protocols can survive their labs counterpart shutting down so you can think long term. - Unified liquidity: spot flows easily even though margin requires deeper cross program APIs; - Answer to some regulatory risk: jurisdictions have a right to regulate their territories but are forced to apply different rules to credibly decentralized systems that span legal systems. This vision hasn’t been fully delivered yet; but on-chain is what unlocks it; quite frankly: the state transitions need to be on distributed consensus; otherwise the admin of state will one day exploit or fall short.

Seoul ✅ Now in Singapore 🇸🇬 At KBW we announced our on-chain verifiable, capital-efficient perp DEX on Solana. Now looking for ~50 experienced perp traders to join the alpha and shape how it gets built. At TOKEN2049? Market makers and perp traders, come find us.

Most Solana programs are one typo - or one agent mistake - away from disaster. Security pros warned us 5 years ago to not do this, but everybody does it. Examples do it. Agents do it. Auditors clear it. Millions drain. fn withdraw(...) -> Result { if record.owner != *user.key { return Err(IllegalOwner); } if amount > record.quote { return Err(InsufficientFunds); } // more `if`s here - hope we have enough! record.quote -= amount; transfer_out(destination, amount) } For both safety and performance, it's time to learn from decades of critical systems engineering, embrace the strength of Rust, and "Parse, Don't Validate." Runtime checks don't belong in a program's internals. They belong at the boundaries. Read Here 👉 👉

Seoul wrap ✅ Solana Summit with @SuperteamKorea. Frontier Traders with @joinfrontier. And a stop at @fomo. Great few days meeting traders, MMs and teams across the @solana ecosystem. Excited for the launch of Flint perps?

Excited to say I have joined the team at @flint_trade_ ! Great leadership, strong team, decentralized vision. Follow along as we question assumptions, push the boundaries and build the most transparent, efficient perp DEX we can dream up. If you're a builder or curious about decentralized exchange ideas, notis on, let's talk this out If you're a market maker, reach out - we have the lowest friction ways to quote spot and soon perps I highly suggest following cofounders @thedavidgorski (ex Jito BAM tech lead) and @Josh_E_Wa (ex DYDX), and shoutout to @SuperteamCAN . I met David at our office meetup.

@flint_trade_ is building perps on @solana. I'm very excited to start sharing more about our goals; especially regarding the level of verifiability, permissionlessness and on-chain logic. My off-the-cuff thoughts in this video.

Flint is building a fully on-chain perp DEX on @solana! Our premise: real DeFi is on-chain and verifiable. No off-chain matching engine. No centralized sequencer. No need to bootstrap our own validator network. More below.

Markout analytics ship with the dashboard. Where you're being adversely selected, your book against top of book, quoter session health and latency, and how you price against other makers on the same flow.

Appreciate all submissions. Congrats to the winners! 🥇 750 USDC — 🥈 500 USDC — 🥉 250 USDC — 353 entries. Thanks to everyone who took the time.

OracleOffset gates each level, so the book hollows from the inside as the fair ages. LinearDistribution uses one backoff rate. Every level stays live, the whole range just steps outward per slot.

One staleness bound for the whole book is a blunt tool. Quote OracleOffset on Flint and each of your 16 levels per side carries its own, in slots. Tight inside, looser outside. As the fair ages the book widens from the inside instead of filling at a stale price.