Bots ateyourlunch.
A sandwich bot buys right before your trade and sells right after it, so you pay more and it keeps the difference. LUNCH finds every time it happened to your wallet, shows what it cost you, and tunes your slippage so it stops.
sandwich attacks detected on Solana in 11 months
different wallets hit by at least one
of attacks landed on pump.fun's curve or PumpSwap
SOL taken by 2,545 bot wallets, before their costs
One block. Three trades.
- 01
You hit buy.
1.00 SOL into a pump.fun coin with slippage at 15%. The quote says 21.46M tokens. Your transaction waits for a spot in the next block.
- 02
A bot gets there first.
It sees your buy and lands 2.00 SOL of the same coin one position ahead of you. The price jumps 10.2%, still inside your 15%.
- 03
You fill at its price.
Your 1.00 SOL now buys 19.49M tokens, 9.2% fewer than quoted. It is inside your slippage, so the trade goes through.
- 04
It sells straight into you.
One position later the bot dumps everything and keeps 0.094 SOL. That came out of your trade.
Illustration on a pump.fun bonding curve holding 10 SOL. Same math LUNCH uses to price your real losses.
Lunch Check
Every swap you made on pump.fun or PumpSwap, checked against the transactions around it in the same block. Each hit links to the three transactions on Solscan, so you can verify it yourself.
RPC endpoint
LUNCH reads public Solana RPC by default. It is free and rate-limited, so scans take a while. Paste your own endpoint (Helius, QuickNode, Triton) to make every tool faster. It is stored in this browser only.
Slippage Tuner
Slippage is how far you let the price move before your trade fails. Set it high and a bot has room to push the price up to that limit. The tuner reads the coin's latest trades and finds the tightest setting that still lands.
High slippage gets eaten.
How often a swap was sandwiched, by the slippage the trader allowed. The safest band was 1–3%. Above 10%, the hit rate climbs fast.
Sandwiched swaps per 10,000 Jupiter swaps routed through pump.fun's curve or PumpSwap, 30 Oct 2025 – 7 Oct 2026. Observational data from Bitquery. Source ↗
Swaps that allowed 20–50% slippage were sandwiched about 18 times as often as swaps at 1–3%. That is 148 versus 8 per 10,000.
Tune my slippageSandwiches as they land.
The monitor pulls the newest Solana block every few seconds, replays every pump.fun and PumpSwap trade inside it, and flags each sandwich it finds. Real blocks, sampled, nothing simulated.
Downloads full blocks, so it uses a few MB per minute. It pauses when you leave the tab.
How LUNCH finds a sandwich.
No guessing and no private data. Everything comes from public Solana transactions and is computed in your browser.
Pull your swaps
LUNCH reads your latest transactions and keeps the ones that bought or sold a token on pump.fun's bonding curve or PumpSwap.
Read the block
For each swap it opens the block it landed in and looks at the transactions just before and just after yours on the same pool.
Match the bot, price the damage
One wallet buying before you and selling after you is a sandwich. LUNCH then replays your trade on the pool's reserves from before the bot moved them, and the gap is what you lost.
Every result shows the front-run, your trade and the back-run as Solscan links. If a number looks wrong, the transactions are one click away.
Run a checkThe token behind the scanner.
$LUNCH funds the tools that keep traders from feeding bots. Phase 1 is live on this page. The next two are being built, and each says plainly where it stands.
Diagnose and fix
- Lunch Check for any wallet
- Slippage Tuner for any pump.fun token
- Live sandwich monitor (beta)
LUNCHBOX
A protected buy button. Trades are sent through Jito with its don't-front marker set, so Jito bundles cannot place a buy ahead of yours. It cuts the attack surface. It cannot block bots that see your trade through other routes, and we will say so on the button.
Lunch money
Creator fees fill a pot that goes back to the holders who lost the most to bots, verified by Lunch Check against on-chain data. Rules, epochs and amounts get published before the first payout.