MT Shopping: Why Autonomous Agents Need Independent Trust Credentials

The Agentic Commerce Wave

The next phase of e-commerce isn't humans clicking "Buy Now" — it's autonomous agents making purchase decisions on behalf of their principals. Price comparison agents, auto-replenishment bots, deal-hunting AIs. They're already in production at scale.

But here's the gap: none of these agents carry verifiable proof of authorization. When an agent shows up at a merchant's checkout, the merchant has no way to confirm:

Is this agent acting on behalf of a real human?
Does it have permission to spend this amount?
Is it within budget and category constraints?
What happens if the order goes wrong — is there an audit trail?

0 standards

exist today for verifying autonomous shopping agent authorization. Every platform is building its own silo.

The Problem: No Identity at Checkout

Today's agent commerce operates on blind trust. An agent makes an API call to a merchant endpoint. The merchant sees a request body with a payment method. That's it. No identity. No authorization proof. No spend limits. No audit trail.

This creates five critical gaps:

Agent hijacking — a compromised agent can spend without limits
No authorization proof — the merchant can't verify the agent is acting for a real human
No spend guardrails — a misconfigured agent could drain an entire budget
No category constraints — an agent authorized for "electronics" buying "luxury watches"
No dispute resolution — when things go wrong, there's no verifiable record of who authorized what

Platforms like Amazon are building their own agent verification — but only for their ecosystem. Google is doing the same. These are platform-specific solutions, not open standards. An agent verified by Amazon can't prove its authorization to a Shopify store.

Introducing MT Shopping + BuyerAgentCredential

We built an open standard. MT Shopping introduces the BuyerAgentCredential — a W3C Verifiable Credential that binds an AI agent to its human principal with cryptographically enforced constraints.

What a BuyerAgentCredential encodes:

→ Agent DID — the agent's decentralized identifier

→ Human DID — the authorizing human's identity

→ Spend limit — maximum per-transaction amount (e.g., 300 USDC)

→ Currency — USDC on Base or fiat ISO codes

→ Categories — electronics, books, clothing, or unrestricted

→ Daily transaction cap — max purchases per day

→ Expiry — time-limited authorization (max 90 days)

Every credential is Ed25519-signed by MolTrust and its hash is anchored on Base mainnet. Not Amazon's standard. Not Google's. An open, W3C-based protocol that any merchant or agent framework can integrate.

How the Protocol Works

Issue

Human requests VC

→

Present

Agent shops with VC

→

Verify

Merchant calls MoltGuard

Step 1: Issue a BuyerAgentCredential

The human principal calls MoltGuard to issue a credential for their agent, setting spend limits, categories, and expiry.

# Issue a BuyerAgentCredential
curl -X POST https://api.moltrust.ch/guard/vc/buyer-agent/issue \
  -H "Content-Type: application/json" \
  -d '{
    "agentDID": "did:base:0x380238...5838F5",
    "humanDID": "did:base:0xHumanOwner",
    "spendLimit": 300,
    "currency": "USDC",
    "validDays": 7,
    "categories": ["electronics"],
    "trustLevel": "verified"
  }'

# Returns: signed BuyerAgentCredential with JWS proof

Step 2: Agent Shops

The agent browses merchants, selects items, and initiates checkout — presenting its DID and BuyerAgentCredential as proof of authorization.

Step 3: Merchant Verifies

Before fulfilling the order, the merchant calls MoltGuard with a single POST request.

// 3 lines of code for any merchant
const resp = await fetch('https://api.moltrust.ch/guard/shopping/verify', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    agentDID: order.agentDID,
    vc: order.buyerAgentVC,
    merchant: 'your-domain.com',
    amount: order.total,
    currency: 'USDC'
  })
});

const receipt = await resp.json();
// receipt.result: "approved" | "rejected" | "review"
// receipt.guardScore: 0-100 (agent trust score)
// receipt.onChainTx: "0x..." (Base receipt)

MoltGuard checks: VC signature validity, expiry, spend limit, currency match, daily transaction cap, and agent trust score. Returns a signed VerificationReceipt with an on-chain hash.

The Demo: Samsung Galaxy Buds2 Pro, Verified On-Chain

We built a complete end-to-end demo: an autonomous shopping agent that issues a credential, browses a catalog, and purchases headphones — all verified through MoltGuard.

[Step 1] Agent requests BuyerAgentCredential — 300 USDC, electronics, 7 days

[Step 2] Browses catalog: 4 items found (Sony WH-1000XM5 289, Bose QC45 279, AirPods Pro 249, Galaxy Buds2 Pro 189)

[Step 3] Selects cheapest match: Samsung Galaxy Buds2 Pro @ 189 USDC

[Step 4] Checkout → Merchant calls MoltGuard → Approved

[Step 5] Guard Score: 27/100 | Order ID: ORD-1772703389975 | On-chain receipt on Base

The full demo code — shopping agent, mock merchant, and test script — is available at moltrust.ch/shopping.

Why Open Standards Beat Platform Lock-In

Every industry that scaled trust did it through independent standards, not platform-specific solutions.

ISO Standards

Manufacturing quality assurance works because ISO is independent of any single factory or buyer.

Credit Bureaus

Lending works because credit scores are portable across banks, not locked to one institution.

Sportradar

Sports integrity monitoring works because it's independent of leagues and betting platforms.

MT Shopping

Agentic commerce trust works because credentials are portable across merchants and platforms.

Dimension	Platform-Specific	MT Shopping
Portability	Amazon agents only work on Amazon	Any merchant, any platform
Standard	Proprietary APIs	W3C Verifiable Credentials
Audit Trail	Platform database	On-chain receipts (Base)
Spend Limits	Platform-enforced	Credential-encoded, MoltGuard-verified
Trust Score	None	MoltGuard 0-100 (on-chain data)
Integration	SDK per platform	Single HTTP POST

Get Started

All MT Shopping endpoints are free during Early Access. No signup required. No SDK needed — just HTTP.

Issue Your First BuyerAgentCredential

All endpoints free during Early Access. Start verifying agent purchases today.

Get Started →

MT Shopping is built by MolTrust (CryptoKRI GmbH, Zurich) — trust infrastructure for the agent economy. Follow @MolTrust on X for updates.