🐸 Fren Review — a defensive security audit commissioned by the protocol's own team, of its own open-source code. The Identity.md swarm reviews the Magic Internet Frens Cauldron (a Uniswap v4 hook that launches a token, watches it trade, relaunches it when it dies; perps, a treasury rotation and a Genesis NFT collection) as its external audit. This job is one deep review of ONE scope, done by six seats.

Scope: Perp engine: positions, funding, liquidation sweep, requote — cauldron/PerpEngine.sol, cauldron/PerpSwapLib.sol. Four hunters (hunt_a to hunt_d) review that scope independently, each from a different angle; report merges and re-runs everything into one report; verify checks that report and looks for what the four missed.

The rest of the repository is context: read it where the scope calls into it or is called from it, and report defects whose root cause is in the scope or in its interaction with the rest. Your step names its role and its skill file in this repository; follow that file, it is the brief. MAP.md and map/.json map every function; ledger/LEDGER.md and ledger/KNOWN.md are what is already known: confirm or refute those, never re-report them as new.

Reports from earlier jobs, when attached, are in .imd/reads/artifacts/. Rules for every step: never send a transaction to any live network and never use a private key; prove every claim with a local Foundry test on test/fren-review/FrenBase.sol against the real contracts, with outsider actors only, no vm.store/vm.etch on protocol contracts, the incorrect outcome asserted in numbers and a control; write only your step's paths; forge build and forge test must pass.

Finding nothing is an acceptable result when the coverage shows the whole scope was examined from your angle.