Frequently asked questions
Direct answers about Opertus privacy, keys, proofs, and transactions.
Short answers for orientation. Each one links to the page with the full detail.
Is Opertus ready for production?
No. Opertus has not been audited and has no production deployment. See project information for what is missing.
What does Opertus keep private?
Note owners, note values, and the link between a commitment and its nullifier. Deposits, withdrawals, recipients, external calls, and timing stay public. The privacy model has the full list.
Can someone with my view key spend my notes?
No. The view key decrypts notes and derives their nullifiers, but spending also requires a separate BabyJubJub spend key and the note-specific data. A leaked view key still exposes private balances and transaction activity.
How do I prove one payment without sharing my view key?
Share a receipt. It holds that note's key and where it was logged, and anyone can check it against the chain. It opens that note only. Either the sender or the recipient can build it from their seed. See selective disclosure.
Can someone with my spend key find my notes?
Not from the chain by using that key alone. Scanning and decryption use the secp256k1 view key. A spend-key holder who also obtains decrypted note data and its spend ephemeral can derive the one-time spend secret. See keys and notes.
Is a commitment an encrypted note?
No. A commitment is a Poseidon fingerprint of a note. The note plaintext is separately encrypted with XChaCha20-Poly1305. The ciphertext is authenticated against the commitment, chain ID, pool address, ephemerals, and view tag.
Why are commitments and nullifiers both public?
Commitments let the pool record notes without learning their openings. Nullifiers let it reject a second spend. Under the intended construction, a public observer cannot determine which commitment produced a nullifier.
Can an old Merkle root be replayed?
Roots remain recognized, but that does not make a note spendable twice. The pool permanently records nullifiers, so the second spend reverts. An old root also cannot prove membership for a note inserted after that root.
How many notes can one proof spend and create?
A proof spends 1, 2, 4, or 8 notes of one token and creates exactly three output notes. Inputs must equal outputs plus the public amount. See proofs and settlement.
Is a zero-value note ignored?
A zero-value output is still a real note with a distinct commitment. Zero-value inputs are used as padding and do not need Merkle membership. Outputs within one call must be distinct.
How are relayers paid?
The wallet makes the third output note payable to the relayer. The amount
remains private inside that note. The pool neither calculates a relayer fee nor
relies on msg.sender for spend authority. See
relayers.
Can a relayer change my withdrawal or external call?
Not while preserving a valid proof. The external-data hash binds the chain, pool, encrypted notes, target, calldata, protocol fee, return list, and clone salt as applicable. A relayer can refuse or delay submission.
Are swaps private?
No. The input and change notes stay private, but the external call is public, including its target, calldata, amounts, and returns. A swap submitted through the public mempool can be sandwiched.
What happens to tokens returned from an external call?
The caller names up to four distinct return tokens and a minimum value for each. The pool measures what actually arrived and creates a note for that value. Transfer-taxed tokens therefore note the received amount, not the nominal amount. An unlisted token can remain on the clone.
What is the difference between salt zero and a nonzero clone salt?
Salt zero uses the shared clone for calldata execution and does not identify a
persistent user clone. A valid nonzero salt is
Poseidon(spendScalar, diversifier) and selects a repeatable CREATE2 clone,
which can support state tied to msg.sender.
Can the pool owner take deposited assets?
The owner powers are pause/unpause, one-time executor setup, delayed fee and treasury changes, and claiming accrued fees. There is no owner-only function to sweep idle pool reserves. Ownership still creates availability and governance risk because pausing is immediate.
What happens when fees change?
Deposit and unshield fees are capped at 10%. Proposed changes wait one day.
Because the unshield protocol fee is proof-bound, a proof built for an old fee
fails after the new fee is applied and must be rebuilt. OpertusClient.withdraw
reads the live rate before proving.
How do I check a deployment?
Follow the deployment integrity checklist in the threat model.