Source: https://trusteed.xyz/en/demo

Live demo · no sign-up

# Try how the rule engine behaves without writing a line of integration.

The demo store's public endpoint answers read tools with no authentication. Run a full scenario step by step, or configure an agent profile and watch which merchant rules fire before the transaction reaches checkout.

live Public endpoint: `POST api.trusteed.xyz/demo-store/mcp/{discovery|checkout}`

Call search\_products or get\_merchant\_profile directly: the read tools need no key. create\_cart, preview\_checkout and complete\_checkout are writes and do need one: without it they answer no\_credentials. Request a sandbox key at POST /api/v1/sandbox/key (3 per IP per day, expires in 24h) and send it in the X-Agent-Api-Key header: this page does that for you, and it works the same from outside. That shortcut exists ONLY on demo-store and the checkout is simulated (this store has no payment gateway connected): a real merchant requires OAuth 2.0 machine-to-machine, a payment mandate and an agent trust signal. Every step shows the JSON-RPC you would send and the response exactly as it arrives.

Here you see what gets blocked and why. Switch any of the thirty rules on or off, move the agent's signals and check the verdict against the specific rule behind it. It is the same family of rules you would apply to your store.

Two minutes: run “Rule engine block” to see an operation stopped before the charge, and “High-value order” to see when the system requires human confirmation. The rest of the page is the technical detail behind those two decisions.

Guided scenarios 

## A full flow, step by step, against the real endpoint.

Every step runs a demo-store tool and shows the unedited response. Checkout tools need a temporary sandbox key first; the page requests it on its own.

**Scenarios** real

**Search and buy** JSON-RPC 2.0

1. `search_products`Search for sneakers
2. `get_product_details`Product details
3. `create_cart`Add to cart _sandbox key_
4. `get_shipping_rates`Fetch shipping rates _sandbox key_
5. `select_shipping_option`Choose standard shipping _sandbox key_
6. `preview_checkout`Preview the order _sandbox key_
7. `complete_checkout`Complete the purchase _sandbox key_

Run the scenario to see the endpoint's response.

Agent Rules 

## Configure the agent. Switch rules on. See who gets through.

30 of the engine's 54 rules, grouped by category and evaluated against the profile you define. The verdict arrives with the rule behind it and its threshold, not with an opaque number.

Reduced demonstration set: 30 of the engine's 54 rules.

**Agent signals** SIMULATED PROFILE

IDENTITY **Verified identity (R001/R009)** Unchecking leaves the agent unidentified and blocks**Valid signature (R002)** Unchecking invalidates the signature and blocks**Agent revoked (R005)** Checking revokes the identity and blocks**Repeatedly blocked (R005)** Three or more identity blocks in seven days**Blocked at other merchants (R007)** Blocked by two or more merchants in 30 days**Provider confidence (R006)** **0.90** Below 0.6 triggers the low-confidence block**Key age, in hours (R004)** **1** Below 24 h adds new-key friction

HISTORY **Completed orders (R010/R021)** **0** None triggers probation and first purchase**Attempt velocity (R003/R011)** **0** Above 3 triggers the repeat-attempt block**Failed checkouts (R011)** **0** Above 3 triggers the repeat-attempt block**Disputes (R009/R024)** **0** Above 2 triggers the dispute-history block**Cancellations (R014)** **0** Above 3 triggers the delivery-risk rule**Refund ratio (R023)** **0.00** Above 0.5 triggers the refund-abuse block

CART AND CHECKOUT **Return-policy mismatch (R013)** Return expectations clash with the policy**Price drift, in basis points (R015)** **0** Above 100 triggers the price-change guard**Discount codes tried (R017)** **0** Above 5 triggers the discount anomaly**Applied discount, in basis points (R017)** **0** Above 5,000 triggers the discount anomaly**Stored value, in cents (R027)** **0** Above the merchant limit triggers the gift-card cap

SHIPPING AND PRODUCT **PO box (R025)** Triggers the sensitive-address guard**Freight forwarder (R025)** Triggers the sensitive-address guard**Subscription or auto-renewal (R026)** A subscription requires consent**Auto-renewal consent (R026)** Unchecking with an active subscription blocks**Business order (R028)** Requires a signed purchase order**Signed purchase order (R028)** Required on business orders**Regulated evidence (R029)** Required when the merchant preset is regulated

**Merchant rules** 30 / 30 active

**`R001.verified-agent-required` Verified Agent Required** Blocks anonymous agents: requires a verifiable identity (DID/kid) on every checkout.**`R002.signature-spoof-block` Signature Spoof Block** Blocks requests whose agent-token JWS signature is invalid or forged.**`R004.new-key-friction` New-Key Friction** Flags agents using a signing key younger than the configured threshold (24 h by default).**`R005.revoked-agent-block` Revoked Agent Block** Blocks agents with a revoked identity, or blocked 3+ times for identity failures in 7 days.**`R006.provider-confidence-tier` Provider Confidence Tier** Requires the trust provider to report confidence above the merchant's threshold (0.6).**`R008.scope-escalation-detection` Scope Escalation Detection** Blocks agents requesting scopes beyond what the merchant authorised.**`R009.agent-verification-required` Agent Verification Required** Stricter identity gate: blocks checkout when the agent id is missing in high-value contexts.**`R003.mandate-boundary-match` Merchant Order Cap and Category Allowlist** The merchant's order cap and allowed categories. It does not read any agent mandate.**`R007.cross-merchant-abuse-signal` Cross-Merchant Abuse Signal** Flags agents blocked by two or more merchants in the last 30 days: shared abuse intelligence.**`R010.new-agent-probation` New Agent Probation** Restricts new agents: blocks until the agent has completed a minimum number of orders.**`R011.repeat-failed-checkout` Repeat Failed Checkout** Blocks agents with more failed attempts than allowed inside a time window (3 in 300 s).**`R017.discount-anomaly-applied` Discount Anomaly Applied** Blocks anomalous discounts already applied to the cart: number of codes tried and total discount.**`R023.refund-abuse-guard` Refund Abuse Guard** Blocks agents whose historical refund ratio is above the threshold (50 % over 90 days).**`R024.dispute-history-guard` Dispute History Guard** Blocks agents with more disputes or chargebacks than allowed in the configured window (2 in 30 days).**`R012.high-risk-category` High-Risk Category Guard** Blocks orders containing categories the merchant flagged as high risk.**`R013.return-policy-guard` Return Policy Guard** Blocks orders where the agent's return expectations clash with the final-sale policy.**`R015.price-change-guard` Price Change Guard** Blocks when the live price drifts further than allowed from the one seen when the cart was built (100 basis points).**`R016.stock-confidence-guard` Stock Confidence Guard** Flags orders where at least one line has stock below the configured minimum.**`R018.cart-composition-guard` Cart Composition Guard** Flags unusual carts: amount far above the merchant's average, too many lines or an excessive quantity of one item.**`R027.gift-card-stored-value` Gift Card / Stored Value Limit** Caps the face value of gift cards or stored-value instruments per transaction.**`R014.delivery-risk-guard` Delivery Risk Guard** Blocks deliveries to high-risk or sanctioned jurisdictions, and agents with excessive post-shipment cancellations.**`R019.country-jurisdiction` Country / Jurisdiction Filter** Restricts checkout to the merchant's allowed countries and blocks barred jurisdictions.**`R025.sensitive-delivery-address` Sensitive Delivery Address** Blocks shipments to PO boxes or freight forwarders when the merchant requires direct residential delivery.**`R021.first-purchase-with-merchant` First Purchase with Merchant** Flags the agent's first order with this particular merchant: it can trigger review or an extra step.**`R022.payment-rail-restriction` Payment Rail Restriction** Restricts checkout to the payment methods the merchant approved and blocks the rest.**`R026.subscription-autorenew-guard` Subscription Auto-Renew Guard** Blocks subscription or auto-renewal purchases when explicit consumer consent is missing.**`R028.b2b-po-guard` B2B Purchase Order Guard** Requires a signed purchase order on business-to-business operations: prevents unauthorised bulk buying through the agentic channel.**`R020.business-hours` Business Hours** Blocks agentic checkout outside the operating hours the merchant configured in its own timezone.**`R029.merchant-preset` Merchant Preset** The merchant's high-level policy: open, balanced, strict or regulated (the last one requires regulated evidence).**`R030.simple-controls` Simple Controls** Universal safety valve: a global maximum amount and an allowed-country list, regardless of every other rule.

**Transaction and policy** ORDER CONTEXT

Order amount Currency EURUSDGBPCart lines Verification level NONEBASICSTANDARDPREMIUMMinimum buyer trust **0.30** 

Source protocol 

**A person is present** Someone can confirm the operation

**Verdict** SIMULATED IN YOUR BROWSER

Configure the agent profile and press Evaluate rules.

This simulator measures the agent. The Trust Score measures the merchant. They are two sides of the same decision: the merchant reads the agent's signals to decide whether to accept the operation, and the agent reads the merchant's to decide where to buy.

[See how the Trust Score is measured ↗](https://trusteed.xyz/en/trust-score)

What this demo is not 

## No payment is real.

The demo store is a public MCP server with a fictional catalogue. Protocols run in sandbox or test mode: no charges are processed and no funds move. Read tools accept twenty requests per minute per IP, and sandbox keys expire after twenty-four hours. If the endpoint does not answer from this browser, the page says so and shows the equivalent curl: the call works the same from your agent or your terminal.

MCP 

### Real

Model Context Protocol: the scenario calls really do go out against the public endpoint.

**In production.**

Rules 

### Simulation

The rule simulator runs in your browser. The real engine runs on the server and its endpoint is not public.

**It shows the mechanism, not any store's configuration.**

Payments 

### Sandbox

No payment protocol moves funds here. The scenario's checkout ends in a test session.

**No charges.**

Next step 

## Connect your real store

The pattern you just ran is the one applied to your catalogue, your policies and your checkout. Rule configuration is scoped to one operation and one metric.

[Merchant **Talk to us** _↗_](https://trusteed.xyz/en/contact?profile=merchant)[Measure **How the Trust Score is measured** _↗_](https://trusteed.xyz/en/trust-score)[Developer **Integrate an agent** _↗_](https://trusteed.xyz/en/for-agents)
