User: "Buy me a $30 book on Amazon"

Agent:
1. ledger_status({ userId: "user_xxx" })
   → { canSpend: true, balance: 5000, verificationStatus: "approved" }

2. ledger_card({
     action: "create",
     purpose: "Buy book on Amazon",
     type: "single",
     maxAmount: 3500
   })
   → { cardId: "card_abc", lastFour: "4242", status: "active" }

3. ledger_card({
     action: "details",
     cardId: "card_abc",
     summary: "Purchasing book on Amazon"
   })
   → { pan: "4111111111114242", cvv: "123", expiry: "12/27", billingAddress: {...} }

4. [Agent navigates to Amazon, adds book to cart, enters card details at checkout]

5. ledger_transactions({ cardId: "card_abc" })
   → { transactions: [{ amount: 2999, merchantName: "AMAZON.COM", status: "completed" }] }

Agent: "Done! I bought the book for $29.99. Your card has been automatically closed."

[During checkout, Amazon shows 3DS verification page]

Agent:
1. ledger_challenges()
   → {
       challenges: [{
         otp: "847293",
         merchantName: "Amazon",
         amount: 2999,
         expiresAt: 1704067200
       }]
     }

Agent: "Amazon needs verification. Enter this code: 847293"

[User enters code or agent auto-fills if allowed]

User: "I want to start using the shopping agent"

Agent:
1. ledger_status({ userId: "user_xxx" })
   → { verificationStatus: "notStarted", balance: 0, canSpend: false }

Agent: "You need to complete identity verification first."

2. ledger_onboarding({ userId: "user_xxx" })
   → { verificationUrl: "https://persona.com/verify/abc123" }

Agent: "Please complete verification here: [link]"

[User completes KYC]

3. ledger_status({ userId: "user_xxx" })
   → { verificationStatus: "approved", balance: 0, canSpend: false }

Agent: "Great, you're verified! Now you need to add funds."

4. ledger_funding({
     action: "crypto",
     userId: "user_xxx",
     chain: "base"
   })
   → { method: "crypto", address: "0x1234...", chain: "base", asset: "USDC" }

Agent: "Send USDC to this address on Base: 0x1234..."

[User sends USDC]

5. ledger_funding({ action: "balance", userId: "user_xxx" })
   → { available: 10000, currency: "USD" }

Agent: "You're all set! You have $100 available to spend."

User: "Create a card I can use for multiple small purchases this week"

Agent:
1. ledger_card({
     action: "create",
     purpose: "Weekly small purchases",
     type: "multi",
     limitPerDay: 5000,
     requireIntent: true
   })
   → { cardId: "card_xyz", lastFour: "8888", type: "multi" }

Agent: "Created a multi-use card with $50/day limit. I'll need to declare intent before each purchase."

[Later...]

User: "Buy me a coffee"

Agent:
2. ledger_intent({
     action: "declare",
     cardId: "card_xyz",
     summary: "Buy coffee at Starbucks",
     expectedAmount: 650
   })
   → { intentId: "int_123", status: "pending" }

3. ledger_card({
     action: "details",
     cardId: "card_xyz",
     summary: "Coffee at Starbucks"
   })
   → { pan: "...", cvv: "...", expiry: "..." }

[Agent completes purchase]

Admin workflow:

1. ledger_users({
     action: "create",
     externalId: "customer-123",
     firstName: "Jane",
     lastName: "Smith",
     email: "[email protected]"
   })
   → { userId: "user_abc" }

2. ledger_agents({
     action: "register",
     externalId: "shopping-agent-v1",
     userId: "user_abc",
     name: "Shopping Assistant",
     description: "Helps Jane buy products online"
   })
   → { agentId: "agent_xyz", status: "active" }

3. [Create agent token via API for the agent to use at runtime]

1. ledger_risk({ action: "get" })
   → { duplicateDetection: {...}, velocityLimit: {...}, ... }

2. ledger_risk({
     action: "preset",
     preset: "moderate"
   })
   → { message: "Applied 'moderate' preset", config: {...} }

3. ledger_risk({
     action: "update",
     velocityMaxPerHour: 30,
     dryRun: true
   })
   → { message: "Config updated", config: {...} }

[Monitor webhooks in dry-run mode, then disable dryRun when ready]

{
  "error": "cardId is required"
}