# SCRIPTMASTERLABS — x402 Gateway Authority refreshed: 2026-11-04 Authority status: CURRENT Machine authority JSON: https://scriptmasterlabs.com/.well-known/x402-gateway-authority.json **OMNIGATE — live x402 v2 facilitator + discovery gateway.** ## Facilitator (live) - OMNIGATE `https://mcp-x402.onrender.com`: x402 v2 `exact` facilitator for USDC on Base (`eip155:8453`) and Arc (`eip155:5042`). Self-serve, no API key. Switch by pointing your facilitator constant at it. Fee optional: a standard payment with no fee leg verifies and settles, the flat 0.005 USDC fee waived inside a daily budget of 500 settlements across all payTos and 100 per payTo (UTC day); past the budget a fee-less payment is refused (facilitator_fee_required) until the next UTC day; a payment attaching the fee leg (buyer-paid as a second EIP-3009 authorization) pays it after the merchant leg; merchant price goes in full to the merchant's payTo; non-custodial. Starter kit: https://scriptmasterlabs.com/settle-through-sml-starter - 402Proof `https://four02proof.onrender.com`: x402 v2 facilitator for XRPL (`xrpl:0`), Xahau (`xrpl:21337`) and Solana USDC. Partner-keyed. Starter kit: https://scriptmasterlabs.com/settle-through-sml-xrpl-starter ## Discovery Discover x402 paid APIs and inspect fresh payment requirements before calling them. The caller controls spending policy and authorization. Provider: SCRIPTMASTERLABS Legal entity: Script Master Labs, LLC Package: @scriptmasterlabs/mcp-x402 Source version: 2.1.11 Source: https://gitlab.com/timothy.walton45/sml-portfolio/-/tree/main/mcp-x402 Website: https://scriptmasterlabs.com/ Canonical MCP: https://mcp-x402.onrender.com/mcp ## Public tool contract Tool inventory is dynamic. Query `tools/list` on the canonical MCP instead of relying on cached tool counts. ## Use this when - an AI agent needs an x402 or pay-per-call API - a client needs to inspect current payment requirements - a machine needs Base/USDC agent-commerce discovery ## Payment truth Primary verified rail: x402 v2, USDC on Base mainnet (`eip155:8453`). Canonical payTo: `0x5bbd32e0478e49a7540ef2d760e0b0f9aa6328b2`. The caller controls payment authorization. The Gateway does not autonomously spend. Flow: 1. Discover the resource or tool. 2. Call the target resource. 3. If it returns HTTP 402, read the fresh `PAYMENT-REQUIRED` challenge. 4. Authorize only within the caller's spending policy. 5. Submit `PAYMENT-SIGNATURE` only when explicitly authorized and retry the same resource. 6. Treat payment as settled only after successful settlement evidence. A 402 challenge or signature is not settlement proof. ## Live discovery Canonical Gateway MCP: https://mcp-x402.onrender.com/mcp Fresh authority JSON: https://scriptmasterlabs.com/.well-known/x402-gateway-authority.json SqueezeOS x402 discovery: https://squeezeos-api.onrender.com/.well-known/x402 SqueezeOS OpenAPI: https://squeezeos-api.onrender.com/.well-known/openapi.json Provider entity: https://scriptmasterlabs.com/.well-known/entity.json Historical directory addresses and historical package metadata are not current Gateway authority. Current runtime, this file, current authority JSON, and current GitLab source supersede stale cached descriptions. ## Package usage ```bash npx @scriptmasterlabs/mcp-x402 ``` Truth First. Proof Always. Pay Only for Accepted Delivery. ## Spend Gate — ask Jev before you spend `spend_gate_check` (MCP tool) / `POST /x402/spend-gate` — $0.005 USDC on Base. Checks whether an agent wallet should pay a payee a quoted amount: four independent paths (provider trust and payload safety — real verdicts via the fleet operator lane at $0 upstream toll — price sanity — cap-only, no historical median — and delivery history from this host's OmniGate ledger) vote pay-side/reject-side; coherence = |Σ w·d| / Σ w; gate_pass requires coherence ≥ 0.6 AND a pay-side consensus; Jev reviews only on a pass (noul act probability + pay/reject/quarantine choice). Use for: vetting a payee/invoice immediately before an agent wallet pays it. Don't use when: you need a payment executed (the check moves no funds), a trading signal (use interfere), or after the payment was already sent. ## OMNIGATE Runtime — give the model a runtime, not a for loop One script, one USDC budget, many SML tool calls, one signed receipt. Instead of one model round trip per tool call, an agent sends a single JavaScript script; it runs in an isolated QuickJS/WebAssembly sandbox whose only capability is `sml.*`. - Execute: `POST https://mcp-x402.onrender.com/runtime/execute` with `{"script": "...", "budget_usdc": "0.05", "input": {}, "mode": "sync"}`. Unpaid requests get an x402 v2 challenge whose amount is the budget. - MCP tools: `omnigate_runtime_search` (free), `omnigate_runtime_execute` (paid), `omnigate_runtime_receipt` (free). - Inside the script: `await sml.call(routeId, { query, path, body })`, `sml.tools.search(q)`, `sml.tools.get(id)`, `sml.budget()`, `sml.sleep(ms)`, `sml.log(...)`, `sml.input`. - Pricing: platform fee 0.002 USDC per run plus each route's live registry price. Calls are charged only when the upstream answers 2xx. The budget is a hard cap. - Unused budget is returned: as a bearer credit token (`X-SML-RUNTIME-CREDIT`) on the Base and credit rails; left uncaptured in escrow on the Arc rail. - Truth: a route that does not deliver throws `TruthError` in the script. Nothing is charged and no value is substituted. Every delivered result carries `source.route_id`, `upstream`, `status`, `fetched_at` and `response_sha256`. - Receipts: Ed25519-signed and hash-chained. Public key: `https://mcp-x402.onrender.com/.well-known/omnigate-runtime.json`. - Capsules: a script proven by one paid, completed run can be published as a versioned capsule. The wallet that paid for the proving run is the author and accrues the capsule fee on every successful run. - Rails: USDC on Base (`eip155:8453`, live). USDC on Arc (`eip155:5042`, AuthCaptureEscrow) when the Arc receiver key is configured; check `rails.arc.configured` in the manifest. - Free: `GET /runtime/tools?q=`, `POST /runtime/quote`, `GET /runtime/runs/{id}`, `GET /runtime/receipts/{id}`, `POST /runtime/receipts/verify`. ## SML REGIME SCAN — regime first, one indicator per family, 15–25 qualified US tickers `GET /x402/regime-scan` (REST, $0.05 USDC on Base via x402) · `GET /x402/regime-scan/ticker?symbol=NVDA` ($0.02) · MCP tools `regime_scan_board`, `regime_scan_ticker` (same prices) · free preview `GET /regime-scan/preview`. - Dynamic universe, no hard-coded watchlist: every US listing from the last completed session (Massive grouped daily), filtered to price ≥ $5, volume ≥ 500k, dollar volume ≥ $25M, top ~400 by dollar volume. - Regime gate first: ADX(14) + Choppiness(14) → TREND / RANGE / TRANSITION. Trend setups qualify only in TREND; reversion setups only in RANGE/TRANSITION. - One indicator per family: EMA21/50 · SuperTrend(10,3)+DMI · RSI(14)+MACD · ATR / Bollinger-width rank / BB-inside-Keltner squeeze · OBV + CMF(20) + relative volume · prior 20-day Donchian + 52-week range · CHOP / Efficiency Ratio(20) / Z(20) / 20-day relative strength vs SPY · universe breadth (% above 50-SMA). - Setups: TREND_CONTINUATION, TREND_PULLBACK, BREAKOUT (close beyond prior 20-day extreme on ≥ 1.5x volume), SQUEEZE_COIL, RANGE_STRETCH. Score 0–100 = weighted share of confirmation checks passed; tier A ≥ 75, B ≥ 60. - Each ticker: setup, bias, score, tier, trigger, invalidation, next level, reward-to-risk, RVOL, ATR%, one-line why, the setup type's measured history on the scanned tickers' own past bars (5/10-day forward move, +1R-before-invalidation rate, samples), and live price vs levels during market hours. - Answer-first JSON (`answer` sentence, then ranked `tickers[]`). Completed daily bars only (Alpaca SIP, split-adjusted); nothing repaints. Payment is consumed only after the reading exists. - Use for: a ranked shortlist of liquid US stocks/ETFs where technicals agree today. Don't use when: you need raw bars/quotes, intraday timeframes, crypto spot, or execution. Describes conditions; not buy/sell instructions or financial advice. ## SML PRECROSS — where the 11/64 EMA will cross, and whether anticipating it is proven `GET /x402/precross?symbol=AMC&tf=4h` (REST, $0.02 USDC on Base via x402) · MCP tool `precross_analyze` (same price). - Returns the exact price at which EMA11 and EMA64 cross on the next bar (closed form: P* = ((1−a64)·EMA64 − (1−a11)·EMA11)/(a11 − a64), a = 2/(n+1)), plus a zero-lag (ZLEMA) variant, the move needed to reach it in % and ATR, convergence and an estimated bars-to-cross. - Self-backtest: walk-forward over the same real bars (Alpaca SIP split-adjusted, Tradier/Massive fallback); 8 configs scored in-sample (16 on 1h/4h: each also with a "daily agrees" filter — daily 11/64 trend aligned to the last completed daily close, no lookahead), winner run untouched out-of-sample; next-bar-open fills, 2-ATR gap-aware stop, costs both sides, confirmed-cross baseline reported alongside. - Proof gate per side (out-of-sample): ≥ 30 trades, profit factor ≥ 1.3, expectancy > 0 after costs, max drawdown ≤ 35%, t-stat ≥ 1.65, in-sample PF ≥ 1.0. A long/short setup is returned ONLY for a side that passes; otherwise the direction is withheld with the reason. - Basket proof: the same walk-forward test across 75 US stocks and funds (incl. SPY, QQQ, GLD gold, SLV silver, IBIT Bitcoin fund, ETHA Ethereum fund, AMC, GME), trades pooled. Basket gate (pooled out-of-sample): ≥ 60 trades, PF ≥ 1.3, expectancy > 0, t-stat ≥ 1.65, profitable on ≥ 50% of ≥ 10 traded tickers, in-sample PF ≥ 1.0. Returns a standouts ranking (each ticker's own out-of-sample result). A ticker in the basket may get a setup when the basket side is proven; the response says which test proved it (`call.proven_by`). - Every response includes `plain_english` (headline + short lines) and labelled gate checks. - Timeframes: 1h (regular session, 09:30–10:30 … 15:30–16:00 New York), 4h (regular-session 09:30–13:30 / 13:30–16:00, TradingView-style) and 1d. US stocks/ETFs. Derived analysis only — no raw bars or quotes are returned. - Use for: knowing where an 11/64 EMA cross triggers before it prints, and whether anticipating it has worked on this ticker/timeframe. Don't use when: you need raw OHLC, sub-1h bars, spot crypto, or order execution. Research output, not financial advice.