Product
<!-- impeccable:product-schema 1 -->Platform
web
Users
Primary (confirmed by the founder): migrant workers abroad sending money home to family, mostly on a phone, often on a weak or intermittent connection, on a repeating rhythm (weekly or monthly). The receiver is a family member who claims the money and cashes out to local currency; they may have no crypto wallet and no XLM, and may open a link sent over chat.
Other confirmed audiences the product serves directly:
- Operators (the corridor's admin/compliance team): watch pool health, custody and reserves, policy per corridor, monitoring signals, admin actions that need the operator key.
- Regulators and auditors: verify selective disclosures, import a view-only note, run travel-rule exchanges, export compliance reports.
- Evaluators right now: hackathon judges and Stellar Community Fund reviewers who open the landing page, the deck, and the live testnet app.
Product Purpose
Tukar is a private cross-border remittance corridor on Stellar. A sender deposits USDC into a shielded pool with zero-knowledge proofs; the receiver claims it privately and cashes out through an anchor to local money. Amounts and links between sender and receiver are hidden on-chain while compliance stays provable: allow-list membership (ASP), per-corridor caps and disclosure policy, travel-rule exchange, proof of reserves, and selective disclosure to a regulator on request. Success is a family receiving the full amount, fast, with a receipt they can show, and a compliance officer able to answer a regulator without seeing everyone's transactions.
Positioning
The only remittance flow on Stellar where privacy is on-chain (Groth16 BN254 proofs verified by
Soroban contracts) and compliance is on-chain and provable too, not a promise in a PDF. On the live
corridor that means an in-circuit allow-list and deny-list checked on every deposit, a per-corridor
policy registry, and four disclosure types each verified by its own Soroban contract. Three further
compliance contracts are deployed and exercised on a preview track that does not touch the live
corridor: the enforcement pool, the proof-of-reserves liability accumulator, and the timelocked
admin. Applying them to the live pool is the first tranche of the SCF build proposal, not something
that has already happened. Proof of reserves is the weakest of the four today: the reserves
contracts point at the live pool but have never been attested against it, so latest_attestation()
returns null and the aggregate contract's covered_count() is 0. The wiring is proven end to end
against test-double pools; on the live corridor it is wiring, not evidence. Reusable KYC instead of
re-verifying every time: Reclaim's zkTLS proof is bound to the wallet address server-side and does
populate the on-chain allow-list, while an idOS credential is verified but cannot be tied to a Stellar
address with a consumer's permissions, so it adds no allow-list entry and the app says so. Everything
described here is deployed on Stellar testnet and verifiable from the explorer, on the live corridor
or on the preview track as labelled above; nothing in the product is mocked.
Operating Context
- Sender: connect a wallet (or the built-in testnet key for demos), pick a corridor (MX, BR, AR, PH, ID, VN, TH, IN, NG, CO), enter an amount, see cost and policy, send; the app proves in the browser (about 20 to 60 s), then hands over a bearer claim note, a claim link (optionally PIN-wrapped), a proof-of-payment link, and a view-only note for a regulator. Sent notes can be cancelled and refunded until claimed. Recurring plans exist (server mode needs provisioning).
- Receiver: paste or open a claim note, reveal the amount, see the live on-chain FX quote (Reflector oracle on covered corridors, HTTP fallback elsewhere), cash out via the anchor (SEP-24 against the SDF testnet reference anchor) or an off-ramp quote, print or share a receipt, optionally put idle USDC into Blend (testnet yield).
- Operator console: pool health, custody vs liabilities, reserves attestation, corridor policies, oracle health, monitoring (deposit velocity, near-cap and repeated-actor heuristics, admin events), contract inventory, admin CLI builders (the admin key never enters the browser).
- Regulator console: verify receipts, selective disclosure (exact, threshold, range, aggregate), audit requests, view-only note import, travel rule (TRP 3.2.1 self-hosted, Notabene sandbox, TRISA node), compliance export (PPATK, EU TFR, BSP presets), trail.
- Public /verify page: paste a receipt or open a verification link; re-verifies on-chain.
- Deck at /deck (HTML slides) and a landing page at /.
- Network: Stellar testnet, Protocol 28. Fiat edges are the SDF reference anchor today; a licensed anchor is a business step before mainnet.
Capabilities and Constraints
- Stack: Next.js 15 App Router, React 19, TypeScript, Tailwind, one app under
webapp/. Contracts in Rust (soroban-sdk v26), circuits in Circom, proving in-browser with snarkjs (WASM, multi-worker). - Real constraints that shape the interface: proving takes tens of seconds on a phone and must not be interrupted; the claim note is a bearer secret (whoever holds it can spend); deposits and withdrawals are public at the edges by design, the crossing in between is shielded; every honest failure state is real (network black-holed, oracle stale, integration not configured) and must read as such.
- Everything must stay honest: no mock data, no fake users, testimonials, or numbers. Integrations that are not configured say so.
- Terminology: note (bearer claim secret), commitment, shielded pool, corridor, ASP allow-list, disclosure, audit request, off-ramp, anchor, relayer.
- Accessibility: keyboard-only flows and screen-reader labels are tested (axe, Playwright, WCAG AA contrast); keep them.
- Copy voice: plain, calm, factual. No hype, no em-dashes in prose.
Brand Commitments
- Name: Tukar (Indonesian for "exchange"). Founder is in Indonesia; the Indonesia and Philippines corridors are emotionally primary even though Mexico has the deepest oracle coverage.
- The incumbent look (dark, orange accent, monospace details) is NOT a commitment: the founder asked for a full redesign with a new visual world. Product truth, copy, and function stay; the old look is evidence and anti-reference only.
- Assets: SVG icon (
webapp/app/icon.svg), og-image (webapp/public/og-image.png), a demo video used by the deck (webapp/public/demo-id.mp4). No other brand assets exist; do not invent logos beyond a wordmark treatment.
Evidence on Hand
- 5th place, Stellar Privacy / Real-World ZK hackathon (DoraHacks); APAC hackathon grand finalist.
- 15 contracts live on Stellar testnet with explorer links (README contract table,
deployments/testnet.json); real deposit transactions on Protocol 28. - Test evidence: 333 contract tests across eight crates, 282 unit tests across 36 files, Playwright multi-browser e2e (chromium, firefox, webkit, 390px mobile), qa6 sweep 66/0.
- No real user testimonials, no volume numbers, no partner logos. Do not fabricate any.
Product Principles
- The family gets the money: speed, clarity, and a receipt beat everything else on consumer surfaces.
- Privacy and compliance are both shown, never asserted: every claim links to something on-chain.
- Honest states are first-class: waiting, failing, not-configured, and "we cannot read the chain" are designed, not hidden.
- The phone on a bad connection is the primary device for senders and receivers; consoles are desktop-first but must still work on a tablet.
- Nothing in the interface is decorative fiction: numbers, badges, and progress come from real state.