OKX.AI Genesis 2026 · Finance x402 exact · EIP-3009 X Layer eip155:196 USD₮0 · zero-gas

Grade the bet against the closing line, not the scoreboard.

You can win the bet and still be wrong. CLV Scout is a micro-priced A2MCP grader: paste a bet you placed — or up to 25 of someone else's — pay a cent or two over OKX x402, and get back a calibrated Closing-Line-Value grade (A+…F). Not an opinion — a number backed by a settled truth table you can independently recompute.

124/124 vitest passing 100% coverage (stmts·branch·funcs·lines) 4-curl self-check: 402·402·405·200 sample audit = real output, unedited
CLV Scout animated hero — a price path beating the dotted closing line inside a scout reticle
01 The paid edge

One request, one cent,
one on-chain receipt

Every paid route is gated by the OKX x402 exact scheme. No account, no API key, no subscription — the HTTP status code is the paywall, and settlement lands on X Layer in USD₮0 with zero gas.

Probe unpaid

POST /api/grade with no payment. The server answers with a real challenge — price, network, asset, pay-to address.

402 · x402Version: 2

Sign & pay

The buyer signs an EIP-3009 TransferWithAuthorization (EIP-712, via viem) for exactly $0.01 in USD₮0 and replays the request.

X-PAYMENT: <signed auth>

Verified, settled, graded

Signature is verified with real offline cryptography, settlement goes to the OKX Facilitator on X Layer eip155:196, and the grade engine runs.

200 + PAYMENT-RESPONSE (tx)

Re-verify any receipt

Anyone can re-check a settlement for free — POST /api/receipts/verify calls the Facilitator's GET /settle/status live.

free · trustless re-check
02 Tool surface

Two paid tools, three free ones

Tiered pricing is two entries in one x402 routes map. Free routes stay free under every configuration — that's a tested invariant, not a promise.

FREE
POST /api/calibration
$0.00

The full grade→outcome truth table plus the exact methodology text — computed live from the settled ledger, not hardcoded prose. This is what makes every paid grade auditable.

FREE
POST /api/me
$0.00

BuyerLens-lite: your own grading history, keyed off your X-PAYMENT signature. Payment identity doubles as login — no accounts anywhere.

FREE
POST /api/receipts/verify
$0.00

Re-check any settlement tx against the OKX Facilitator's GET /settle/status. Receipts stay verifiable without trusting CLV Scout.

GUARD
GET /api/grade
405

Wrong verb, honest answer. The self-check gate is four curls: 402, 402, 405, 200 — reproducible by any judge in ten seconds.

03 Proof, not testimonials

+12% raw ROI.
Sharp Score 40.7/100.

No users to quote yet — this is a hackathon build, so here are signals you can recompute instead. A tout posts his last 5 World Cup moneyline picks: 3 wins, 2 losses, +12.0% raw ROI on flat stakes — looks like someone worth following. Feed the exact same 5 picks to /api/audit ($0.20) against the real, settled closing-line ledger:

124 / 124vitest passing · fully offline
100%coverage — stmts · branch · funcs · lines (v8)
402·402·405·200the 4-curl self-check any judge can run
What the leaderboard shows
+12.0% ROI
3–2 on flat 1-unit stakes. Raw P&L says: follow this guy.
What the closing line says
40.7 / 100
Only 1 of 5 picks beat the close. Weighted expectancy: −22.14%. The numbers that predict future results say otherwise.
POST /api/audit → 200  ·  SAMPLE OUTPUT — real, unmodified response from this build
{
  "graded": 5,
  "beat_close_rate": 0.2,
  "grade_distribution": { "A+": 0, "A": 1, "B": 0, "C": 1, "D": 2, "F": 1 },
  "weighted_expectancy_pct": -22.14,
  "sharp_score": { "value": 40.7 },
  "verdict_line": "Sharp Score 40.7/100 — the record may look fine on raw P&L, but the numbers that predict future results say otherwise."
}

Every number above is genuine output from this build against the 24-row settled seed ledger — nothing scripted or hand-typed. The Sharp Score recomputes from its published sub-scores (tested invariant), and the grade bands are fixed round-number CLV cutoffs decided up front, not curve-fit to the seed data.

04 Sponsor defense

Why ONLY OKX

The grader is portable. The economics are OKX-specific — take OKX out and the product dies at the till.

$0.01 with real settlement

The SDK turns a one-cent USD string into atomic USD₮0 and the Facilitator settles it on X Layer per call. On a general-purpose EVM rail, a $0.01 payment is un-economic the moment anyone pays gas.

Zero-gas USD₮0 promo

The USD₮0 zero-gas promo (0x779ded0c…3736) makes the buyer's marginal cost exactly one cent — the only thing that makes "audit a tout's last 25 picks for $0.20" a rational agent action.

Tiered pricing = config

"$0.01" grade and "$0.20" audit are two entries in the same x402 routes map. Multi-route pricing costs a config block in one middleware — not an integration.

Public sold counts

The marketplace shows sold counts publicly, so the 1¢-volume thesis is verifiable by judges without trusting us.

Shared v2 modules

Same exact/EIP-3009 wire as the primary listing, plus X-PAYMENT-identity BuyerLens and Facilitator GET /settle/status receipt re-verification.

A road past one cent

Tier-2 grade-stream targets the aggr_deferred batch scheme: Session-Key signing, TEE aggregation, one on-chain tx for tens of calls a minute at $0.005 each — a stack that doesn't exist elsewhere.

Bottom line: no other micro-payment rail clears $0.01 profitably off a zero-gas promo, and there's no Session-Key + TEE aggregation stack for sub-cent streaming elsewhere. The grader is portable — the economics are OKX-specific.
05 Try it yourself

Run the interactive demo locally

This page is static (GitHub Pages) — it cannot call the API. Nothing here is wired to a fake fetch. The real interactive demo — live 402 challenges, real EIP-3009 signing, the +12%-collapses-to-40.7 reveal — runs on your machine in about a minute. Node.js ≥ 20 and npm are all you need; no account, no faucet, no keys.
terminal — clone & boot
# git clone https://github.com/edycutjong/clvscout && cd clvscout
npm install
npm run settle          # CLV% + grade per row + truth table -> fixtures/ledger-settled.json
npm test                # 124 vitest, fully offline
npm run api             # boots on :4021

Then open http://localhost:4021/ — a served visible proof page (web/index.html) that drives the live paid API. Every button fires a real x402 round-trip: probe → 402 → sign EIP-3009 → pay → 200 — signed server-side with a throwaway key, because browsers can't sign. Nothing on that page is mocked either.

terminal — the listing-gate self-check
curl -i -X POST http://localhost:4021/api/grade        # -> 402, x402Version:2
curl -i -X POST http://localhost:4021/api/audit        # -> 402, x402Version:2
curl -i -X GET  http://localhost:4021/api/grade        # -> 405
curl -i -X POST http://localhost:4021/api/calibration  # -> 200, live truth table

Every paid route answers an unpaid curl with a real 402 challenge. npm run readiness boots an ephemeral instance and runs this exact check (plus truth-table sum invariants) for you.

terminal — scripts/buyer.ts does the whole x402 dance
npm run buyer POST /api/grade '{"match":"NED vs URU","selection":"Netherlands to advance","odds_taken":1.88}'

The buyer script probes the route unpaid, reads the 402 challenge, signs a real EIP-3009 TransferWithAuthorization locally with a throwaway key, and replays with X-PAYMENT. Signature verification on the server is real offline cryptography (viem's recoverTypedDataAddress) — with OKX Facilitator keys set, settlement goes live to X Layer; without them, it's recorded as an honestly-labeled local-pending receipt.

06 Questions

FAQ

Short, honest answers — the same ones a judge would get.

CLV compares the odds you took against the market's final (closing) price. Beating the close consistently is one of the few widely accepted predictors of long-term betting skill; win/lose on any small sample is mostly noise. That's the whole product: you can win the bet and still be wrong — CLV Scout grades the decision, not the outcome.

No — and it doesn't pretend to be. This is a static GitHub Pages site with zero fetches, so it cannot run the API. The real interactive demo runs locally: npm run api boots the server on :4021 with a visible proof page that fires genuine x402 round-trips. Everything shown here (the 40.7 Sharp Score, the audit JSON) is real, unmodified output from that build.

$0.01 per single-bet grade (/api/grade), $0.20 per audit of up to 25 bets (/api/audit). Payment is the OKX x402 exact scheme: your agent signs an EIP-3009 TransferWithAuthorization in USD₮0 and replays the request with an X-PAYMENT header. Settlement lands on X Layer (eip155:196) under the zero-gas promo, so the marginal cost really is one cent. Calibration, /api/me and receipt verification are free.

No accounts anywhere. Your X-PAYMENT signature is your identity — /api/me returns your grading history keyed off it. USD₮0's zero-gas promo means no native-token balance is needed to pay. The local demo signs with a throwaway key, so judges need Node.js ≥ 20 and nothing else.

You get UNGRADED with an explicit reason — never an invented number. The current build covers World Cup 2026 knockout markets by design (a narrow, fully settled ledger beats a wide, guessed one), and every calibration response states that coverage plainly.

Yes. POST /api/calibration is free and returns the full grade→outcome truth table plus the methodology text, computed live from the settled ledger. Grade bands are fixed round-number CLV cutoffs decided up front (not curve-fit), and the Sharp Score recomputes from its published sub-scores — both are tested invariants in the 124-test suite.

Free and trustless: POST /api/receipts/verify re-checks any settlement tx against the OKX Facilitator's GET /settle/status live. Running locally without Facilitator keys, payments are recorded as honestly-labeled local-pending receipts — the build never fakes a settlement it didn't make.

Yes — the JSON above is a genuine, unedited /api/audit response from this exact build run against its 24-row settled seed ledger. Five picks that show +12.0% raw ROI grade out to a 40.7/100 Sharp Score with −22.14% weighted expectancy. You can reproduce it end-to-end with npm run settle && npm run api and the buyer script.

You can win the bet and still be wrong.

Grade the decision for a cent. Audit the tout for twenty. Recompute everything for free.