TukarPayments & Consumer AppsTestnet
Tukar

Send money home privately. Cash out in local fiat.

USDC in · private crossing · local fiat out

A mobile-first cross-border remittance app on Stellar. A worker deposits USDC, the crossing hides the amount and both people, the family cashes out in local fiat. What sets it apart from every other privacy app: it stays provable to a regulator. Private in the middle, accountable at the edges.

Compliance clearedPrivate in transit
TukarProblem → solution02

A public ledger leaks who moved the money.
Tukar keeps the crossing private.

Cross-border remittance is Stellar's core use case, but on a public ledger every payment's amount and both counterparties are exposed. A corridor has to hide the parties too, while staying auditable enough for a regulator.

01 · PUBLIC
Deposit
Real testnet USDC enters the pool. A ZK compliance proof shows the authenticated depositor is allow-listed.
02 · PRIVATE
Shielded crossing
The transfer hides the amount and both counterparties on-chain. Only a commitment and nullifiers are visible.
03 · PUBLIC
Off-ramp
The receiver reveals only at the edge and withdraws to local fiat (IDR, PHP, VND, THB, INR, MXN, BRL, ARS, NGN, COP).
04 · ON DEMAND
Disclose
A holder proves one fact to a regulator (e.g. the amount), with no keys and no other payments.

Who it's for: the migrant worker sending money home and the family receiving it in local currency, plus the licensed anchors and PSPs that pay for a private corridor they can still audit. Public at the edges by design, private exactly where a corridor must not leak.

TukarThe moat03

Compliant privacy on cross-border money. No catalog equivalent.

The category already has privacy tools and consumer wallets. None sits where Tukar does: private for the user and provable to a regulator, on a real remittance corridor.

Compliance is in the circuit

vs a privacy pool / mixer

A mixer is private but not accountable. Every Tukar deposit proves in-circuit that the authenticated depositor is ASP allow-listed and not deny-listed, key pinned to from. Compliance is in the proof, not a promise.

A real cross-border corridor

vs a private payment link

Payment links move USDC between crypto users. Tukar takes real fiat in and out through live SEP anchors on Stellar's rail, 10 corridors, with an on-chain Reflector min-receive gate on the off-ramp rate.

Private, and it verifies

vs a plain consumer wallet

Wallets are cross-border but public. Tukar hides the amount and both parties on-chain, and a regulator can verify a disclosed fact live; tamper with it and the Stellar verifier returns InvalidProof.

The layer anchors plug into

The combination

Each rival covers one axis. Holding all three at once has no off-the-shelf equivalent, so Tukar isn't another privacy app, it's the compliant settlement layer a licensed anchor plugs into, one policy per corridor.

TukarThe opportunity04

A $669B market that still pays over 6% to move.

Cross-border money is huge, slow, and expensive. Crypto rails made it cheap but public, and that trade is exactly what regulated money can't take. Tukar closes the gap.

$669B a year

The market

Remittances into low- and middle-income countries in 2023 (World Bank). Sending $200 still averages about 6.2%, more than double the UN's 3% target, and it has barely moved in a decade.

Cheap or compliant, not both

Why it's still open

Public stablecoin rails cut the fee but expose every amount and counterparty. Privacy tools hide it but can't answer a regulator. Neither is usable for a licensed corridor.

A rail, not another app to acquire

Business model

Tukar is the private settlement leg between anchors. Revenue is a thin take-rate on settlement volume, paid by the anchors and PSPs that route through it. B2B2C, so we don't buy consumers one at a time.

One corridor, one anchor, first

Go to market

Launch a single high-volume lane with one licensed anchor, then add corridors. The post-hackathon path is a Stellar Community Fund build award on the same rail.

Sources: World Bank Migration and Development Brief 39 (Dec 2023); World Bank Remittance Prices Worldwide (avg. cost of sending $200, 2023).

TukarZK is load-bearing05

Seven circuits, eight contracts, doing the actual work.

transfer

Circom · Groth16 · BN254

Note ownership, correct nullifiers (no double-spend), Merkle inclusion, value conservation.

compliance

Circom · Groth16 · BN254

The authenticated depositor is in the ASP allow-list and not deny-listed, key pinned to from.

disclosure ×4

Circom · Groth16 · BN254

One commitment proves an exact amount, a threshold, an aggregate cap across a set, or a two-sided range. Aggregate completeness is enforced on-chain by an audit-request registry.

merkleUpdate

Circom · Groth16 · BN254

Inserting a leaf into a known old_root yields exactly new_root, a trustless tree with no admin backdoor.

All seven are proved client-side in the browser (WASM) and verified on-chain. Eight Soroban contracts live on Stellar testnet (a pool plus seven BN254 verifiers) on Protocol 28, using the BN254 host functions introduced in Protocol 25 / 26. Every one is exercised, down to a double-spend rejected on-chain. Full contract IDs on the closing slide and in the repo.

TukarReal, and verified live06

Not asserted. Verified on-chain.

Oracle-gated settlement

Reflector · SEP-40 FX oracle

The off-ramp rate is read on-chain by the pool, priced on the median of 5 records so one manipulated spot can't move it. The withdraw carries a min-receive gate: the pool re-reads Reflector at settlement and refuses to release below ~99% of the quote, failing closed if the feed stalls.

Real USDC, real ceremony

Real assets · real setup

The pool custodies real testnet USDC: deposit moves the amount you type, withdraw releases it, and the commitment binds to exactly that amount. All seven proving keys come from a real multi-party trusted setup (Hermez Powers-of-Tau), independently verifiable with snarkjs zkey verify.

Genuine passes, tampered fails

On-chain, not asserted

A genuine disclosure verifies against the live Stellar verifier. Tamper with the disclosed amount and it returns InvalidProof. A cross-wallet double-spend is rejected on-chain as NullifierUsed.

TukarDepth of integration07

Load-bearing, not name-checked.

BN254 host fns · P25/26
Load-bearing. All seven Groth16 proofs verified on-chain (pairing checks).
Soroban pool contract
Load-bearing. Real USDC escrow, root · nullifier · commitment sets, ASP policy, audit-request registry, full deposit and withdraw lifecycle.
Circle USDC (SAC)
Load-bearing. The real escrow and settlement asset, not a mock token.
ASP compliance (allow + deny)
Load-bearing. Proven in-circuit and pinned to the depositor's key, so you can't deposit as someone else.
Nullifiers + Merkle root
Load-bearing. Trustless root advance proven in ZK; a double-spend is rejected on-chain.
Selective disclosure ×4
Load-bearing. Prove one hidden fact to a regulator, verified on-chain; aggregate completeness enforced by an on-chain registry.
Reflector oracle · SEP-40
Consumed at settlement as a min-receive price circuit-breaker (median of five, fails closed).
Freighter + in-browser proving
Load-bearing. Real signing, and snarkjs proving client-side so secrets never leave the device.
SEP-24 / SEP-1 anchors
Integrated against SDF's reference anchor; a licensed KYC anchor is the honest production step.

The ask: one licensed-anchor pilot on a high-volume corridor, and a Stellar Community Fund build award, to take Tukar from testnet to the live, regulated settlement layer anchors plug into.

Shipped this cycle: a real OpenVASP TRP 3.2.1 Travel Rule exchange with a TRISA companion node, an on-chain per-corridor policy registry with a preview enforcement pool, Reclaim proof-of-personhood feeding the ASP allow-list, bidirectional Circle CCTP V2, recurring on-chain sends, and cryptographic proof-of-reserves. Still ahead: full-pool live proof-of-reserves, per-corridor enforcement on the live pool (a state migration), and TRISA activation (the operator's VASP cert and host).

TukarLive on Stellar testnet08

The whole flow, on-chain, in three minutes.

Real testnet USDC: deposit → shielded crossing → the local figure read on-chain from Reflector → on-chain withdraw → a disclosure proof verified on-chain → a tampered claim rejected on-chain → the regulator and operator consoles. Every frame is the running app, captured by Playwright. Narrated, with captions.

Short on time? The 1:45 live demo cut is the same capture, trimmed to the payment itself: send, arrive, prove. The PowerPoint deck, in Bahasa Indonesia with that cut embedded, is TUKAR_SCF_DECK.pptx.

TukarTry it, no installTestnet

Real-world money.
Private. Kept compliant.
On Stellar.

One click activates a real testnet key. Send builds proofs in your browser and deposits real USDC; the receiver off-ramps to fiat; a regulator audit is verified on-chain, and a false claim is rejected on-chain.

Recognition: placed 5th in the Stellar Real-World ZK hackathon, and a Payments & Consumer Grand Finalist in the Stellar APAC hackathon.

Honest scope: live on testnet, not yet audited, and the fiat edges run SEP-10/24 against SDF's testnet reference anchor (a licensed anchor is pending for mainnet). Everything in the middle is real and verifiable on-chain right now.

Tukar · USDC in · private crossing · local fiat out
Built with: Circom · Groth16 · snarkjs (BN254) · Poseidon · Soroban / Rust · Stellar Protocol 28 · Reflector SEP-40 oracle · SEP-1 / 24 / 41 anchors · Next.js · TypeScript · Tailwind · Freighter