Give it a Monad address or paste the contract. A board of AI agents reads the chain, argues over the code, and writes the surviving verdict onchain — checkable by anyone, forever.
Every Auditora verdict is attested onchain, bound to the contract's codehash at audit time. Look up any Monad address — no account, no trust in us required.
Four moves take a bare address to a verdict you can actually weigh.
Before a model sees the code, Auditora reads the live chain: who controls it, what it holds, and — for a proxy — where the real logic lives. Context decides whether a bug is theory or a live drain.
The real code plus that on-chain context goes to the Auditor, which proposes candidate findings as structured claims — every one of them still just an accusation, not yet a verdict.
An adversarial Challenger attacks every claim — upholding real bugs, disputing the unproven, and rejecting false positives. A deterministic Judge sets the verdict from what survived.
The verdict is hashed and attested to the Auditora registry on Monad, bound to the contract's codehash — if the code ever changes, the badge goes stale automatically.
Auditora is a first-pass triage layer, not a substitute for a professional audit. Surviving an adversarial challenge is a strong signal, not a proof of safety — models can still share blind spots. An attestation records exactly what the board concluded and when — it is not a certificate of safety. Auditora is loudest exactly where it should be: on the findings that survive attack.