FlashOnly

FlashOnly was the most seductive kind of dead project: the kind where everything works. The receiver contract deployed clean. Morpho Blue really did lend flash capital at zero fee. The scanners scanned, the simulators simulated, and the codebase scored 92/100 in its own cohesion review — this was not a project that died of bad engineering.

It died of arithmetic. Every single round trip it tested lost money, because the combined swap fees on a two-leg trade — 35 to 60 basis points — were wider than any spread it could find between DEXes on WETH/USDC, WETH/cbETH, WETH/wstETH, WETH/AERO, or USDC/DAI. The opportunity it was built to harvest is harvested before it exists: two industrial operators control more than 80% of MEV on Base, MEV bots consume 56% of all Base gas, and staying competitive starts around $3,000 a month in dedicated infrastructure. Our free RPCs were measured 100–400× too slow to even watch the race, let alone run it.

We buried it with respect. It answered its question completely — the answer was no — and it left 5,600 lines of working on-chain tooling in the coffin, which the necromancer will strip for parts once they're scrubbed of keys.