How Magneta Finance is tested, what has been verified, and — just as importantly — what has not been. We would rather tell you the limits than let you assume protections that are not there.
Magneta's contracts have not been audited by an external security firm. There is no insurance fund and no funded bug bounty program. Treat the protocol accordingly: use amounts you can afford to lose, and read the code — it is public.
A third-party audit is planned before the protocol carries significant third-party liquidity. This page will be updated with the firm, scope, date and the full report when that happens — not before.
Every contract is reviewed by a panel of independent AI models, each playing a different specialist role, arbitrated by a separate judge model. The same code is scanned repeatedly — one pass surfaces only about half of the high-severity findings, so the union across runs is what we act on, with a per-finding confidence score.
Static analysis (Slither, Mythril, Aderyn, Semgrep), invariant replay and mutation testing run alongside the panel.
Findings are not taken on trust. The contract is deployed for real on a testnet and the reported attack is executed against it from an unprivileged account. A finding is only treated as real if value actually moves; if the guard holds, the finding is discarded.
This cuts both ways, and has: findings rated critical by reading alone have been disproven on-chain, and rare findings that a single review missed have been confirmed.
Found something? Email info@magneta.finance with the contract address, chain, and the steps to reproduce.
We do not currently run a paid bounty. We will acknowledge every report, and credit you publicly if you want the credit.
