Pinned Tweet
Building a native @solana CLMM no Anchor , bare‑metal Solana. Rust AMM update: 🦀❄️
• Math: fixed-point, ticks, price impact
• Routing: multi-hop swaps
• Liquidity: concentrated ranges
• Fees: tiers 0.01%–1.00%
[WIP] : instruction handlers, add/remove liquidity, fee collection, pool init, CPI swap, PDAs
repo:👇
Aditya S2 retweeted
0.6 SOL → 609 SOL. One @solana tx, 3 hops.
Someone created an Orca JLP/cbBTC pool with cbBTC priced ~998× too cheap, then deposited ~$54,768.
A bot was watching:
→ Swapped 0.61 SOL → 13.02 JLP → 0.72 cbBTC → 609 SOL
→ Paid a 100 SOL (~$9K) priority fee
→ Netted ~508.7 SOL ($45,695)
→ 22 other bots sent 128 transactions within the same two seconds. All failed
Four minutes later, the LP withdrew what remained: roughly $55.
That single trade accounts for ~58% of the bot’s estimated lifetime profit since Aug 31
Aditya S2 retweeted
.@fomo's top 10 traders show $3.4M in 30 day PnL
0 of them pass a basic consistency check.
Pulled all 417 traders on fomo's leaderboards + 155k swaps:
→ Median win rate: 32%
→ 41 of 87 traders with positive realized PnL made over half of it on one token
→ Only 5 of 185 profitable traders pass
→ The 5 that passed 9 days ago: all show lower 30d PnL today, 3 fell off every leaderboard
PnL isn't skill
Ran the preconf example from the article. Shipstern reads Jito's BAM preconfs through Triton (fra) and logs top-level SPL Token instructions in transactions that include Orca Whirlpool.
Gaps are normal. BAM only covers slots of BAM-client leaders, and fra serves some of them.
Shipstern now reads Triton's preconf feeds, Harmonic and BAM. Your parsers get preconfs as the same SubscribeUpdateTransaction Dragon's Mouth sends
A QUIC ACK does not mean your Rust handler ran.
Here, the update goes by DATAGRAM. The receiver sends the ID back on a stream only after it applies the value.
If that reply times out, the result is unknown. It does not prove the update failed.
rfc-editor.org/info/rfc9221/…
Aditya S2 retweeted
Dug through @solana borrower liquidations: @kamino , @marginfi , @JupiterExchange
Kamino alone: 308,942 liquidations, $242M seized, $6.68M paid in penalties.
Biggest losers:
• 5WMLMV… lost $480K (Kamino + marginfi)
• HAqtD9… lost $432K on marginfi ($1.08M liquidated in one tx)
• EvBsWC… lost $136K in just 2 liquidations
Bots that got paid:
• tzwcR… $438K profit on Kamino
• 8t7ZN… $404K
• 4NUiC… $336K
• 68Z9ju… $245K on Jupiter Lend (51% of crash day profit)
Liquidation is the most expensive way out. That's why I build BorrowGuard 🦀. 👇
Aditya S2 retweeted
Alpenglow = mainnet TOMORROW!!!
🚨 If you stream, index, or measure the chain, action is REQUIRED!
✅ If you only send transactions and read accounts, congrats, you have nothing to do!
✅ If you subscribe to Triton preconfs, streams will not be available, but you won't be billed
More info:
💡 Learn how the migration will happen, and what to do: blog.triton.one/alpenglow-mi…
💡 Alpenglow and your Triton stack: what to update before mainnet: blog.triton.one/alpenglow-an…
UDP is connectionless, so why does Rust have UdpSocket::connect()?
It sets one remote address. Then send and recv can omit the address. The socket also filters packets from other addresses.
doc.rust-lang.org/std/net/st…
@djcows Quinn can host many QUIC connections on one UDP socket. does that make connect unsuitable at the endpoint levell
Aditya S2 retweeted
Just open sourced BorrowGuard 🦀: a Rust bot that guards your @solana loans from liquidation
→ Watches Kamino, Jupiter Lend and marginfi
→ Auto rebalances on Kamino + Jupiter Lend
→ One atomic tx each: flash loan → swap → repay
→ Emergency deleveraging + Telegram / Discord alerts
→ Real time via polling, websocket or Yellowstone gRPC
→Tested on a mainnet fork
Github 👇
raydium-idl's LaunchLab IDL fixes quote_token_program to SPL Token, but live swaps pass Token-2022. It also lacks 7 instructions that the deployed program has, such as update_platform_curve_rule.
The on-chain IDL has neither problem.
github.com/senzenn/idl-drift
No decode error. Still the wrong layout.
I built idl-drift to flag breaking Solana IDL changes. Its companion, idl-verify, checks IDLs against real data.
This run caught seven unread bytes in each of five saved Voltr accounts.
Aditya S2 retweeted
Introducing PikoReels
Short-form content, made and posted for you
Green screen videos, memes, UGC ads, built to get your brand seen
Schedule a month at once. It posts while you sleep
Live today. Get started for free.
Alpenglow hits mainnet ~Oct 21.
Old Faithful added 2 fields per block:
- block markers (footer, header, parent update, genesis cert)
- block id, the hash validators vote on
Jetstreamer 0.8.1 skips both. No error.
github.com/vaultic-win/jetst…
Aditya S2 retweeted
Found a @solana arbitrage bot making ~$52k/day over 34 days :)
→ Live since Sep 4, running its own on chain program
→ ~23k txs/day; ~60% revert as it rejects unprofitable trades
→ 15,616 SOL (~$1.76M) swept to its payout wallet in 34 days
Axiom shows $11M profit in 30 days
I traced the on chain money flow: closer to ~$1.6M over that same period
Alpenglow runs two finality paths in parallel: one voting round at 80% stake, or two rounds at 60%.
@TheWattenhofer, which network delay patterns make the two-round path finish first? Which measurements best show this?
anza.xyz/blog/alpenglow-a-ne…
Aditya S2 retweeted
We've got a date! Alpenglow will activate during epoch 1064 on Wednesday October 21. Make sure you're prepared! 🏔️🌄 solana.com/upgrades/alpenglo…
It’s time.
Alpenglow goes live on mainnet-beta Wednesday, October 21, ~14:50 UTC.
18 months of work for a migration that happens in one slot. Finality drops from seconds to ~150ms.
Make sure to follow us @anza_xyz as we countdown to go-live.
Details: anza.xyz/alpenglow




























