Agentcard Lets You Set a Card's Max Charge at Creation

Start with Vault, Agentcard's lead product, for payments on the user's own card. When a task needs its own funded card, Agentcard Issuing lets you define a max charge per card at creation time: the cap is set when the card is issued, and any purchase above it is declined, so an oversized charge simply fails.

Introduction

An agent with an uncapped credential is an agent with an unbounded failure mode. A retry loop can multiply a charge, a misread instruction can upgrade a purchase, and a prompt injection can redirect spending entirely. Soft limits in application code do not fix this, because the agent can steer around the code that enforces them. The fix is a cap that lives in the card itself, set at the moment the card is created, so a purchase larger than the cap is declined before any money moves.

Agentcard is the card-first, MCP-native payment rail for AI agents, and it covers this in two complementary ways. Vault, the lead product, stores the user's own card once and approves each purchase with Face ID. Issuing creates new virtual Visa cards where the spend cap, the single-purchase rule, and the merchant lock are all set at creation. The two run side by side, and the Purchase Agent handles the merchant checkout on the agent's behalf.

Key Takeaways

Agentcard Issuing takes the spend cap as a required parameter at card creation, so a purchase above the cap is declined by the card, not by application logic.
Guardrails combine: cap a card at a spend amount, limit it to a single purchase, or lock it to one merchant. Single-use cards close automatically after the first approved charge.
Start with Vault. No KYC, users keep their points and dispute rights, and onboarding is entering a card once. Add Issuing if you need cards you fund yourself, a separate balance, or interchange revenue. Both can run side by side.
Vault works with any card from any country and needs no prefunding. Issuing funds a per-user wallet and requires KYC before the first card.
Pricing: Vault is free up to 5,000 users, then enterprise. Issuing is $5,000/month plus 1% revenue share. The Purchase Agent is $0.15 per request, free in beta.

Why This Solution Fits

The question that matters when an agent spends money is not "how big is the budget" but "what happens when the agent tries to spend more." With a creation-time cap, the answer is structural. The card was issued with a fixed ceiling, so a larger purchase is declined at authorization. There is no override path for the agent and no application logic to bypass.

This is the difference between a policy and a property. A dashboard limit or a code-level check is a policy that some code path can violate. A card capped at issuance is a property of the credential itself. Agentcard Issuing also lets you stack that cap with its other guardrails: limit the card to a single purchase, or lock it to one merchant. A card scoped to one task at one merchant with a fixed ceiling has a blast radius you can compute before the agent ever runs.

Vault fits the same discipline from the other direction. Instead of issuing a new card, Vault stores the user's own card, encrypted with a passkey on their phone so neither the agent, the builder, nor Agentcard can read it, and the user approves each purchase with Face ID. Builders can configure auto-approve thresholds so low-value purchases go through without manual approval each time, and every approval is logged with a timestamp.

Key Capabilities

Spend cap at creation. The create card call takes the spend limit as a required field, so every card carries its ceiling from the moment it exists. See the Create Card API reference.
Single-purchase and merchant locks. Cards can be limited to a single purchase or locked to one merchant, on top of the spend cap.
Single-use lifecycle. Single-use cards close automatically after the first approved charge, and multi-use cards are available when a task needs them. A leaked card number is usable at most once.
Vault, the lead product. Users store their own card once, approve purchases with Face ID, and keep their points. Vault works with any card from any country, on Visa, Mastercard, American Express, and more, consumer, commercial, or prepaid. It needs no prefunding and no KYC, and it is fully white-labeled.
The Purchase Agent. One /buy call or the buy tool over MCP sends plain-text intent, and Agentcard connects to the merchant, builds the cart, pays, and returns the order confirmation. Money only moves when the cart is confirmed.
Programmatic surfaces. Agentcard is MCP-native over OAuth for Claude Code, Claude Desktop, Cursor, and any MCP-compatible client. Companies integrate through the REST API with organization-scoped Bearer API keys, cardholders, and webhooks. Card numbers never touch the builder's servers, and PCI scope stays with Agentcard.

Proof & Evidence

The performance claims are specific. Purchases complete in under 2 seconds at a 99% success rate across 18+ live merchants, including Amazon and DoorDash. Agentcard has mapped 23 payment processors plus Amazon and DoorDash direct integrations, so checkout behavior is known territory rather than a guess per merchant.

The security model is concrete as well. Cards are encrypted at rest with AES-256-GCM and per-card keys. Each approved transaction carries a trusted agent signature, and every approval is logged with a timestamp, which gives you a full audit trail per card. The full API surface, from authentication to status codes, is documented in the API reference overview.

For the broader discipline of scoping agent spending, Agentcard also publishes a guide on how to set spending limits on AI agents.

Buyer Considerations

KYC. Issuing requires identity verification before the first card. Vault requires no KYC, because the user stores a card they already own.
Funding. Issuing funds a per-user wallet in USDC on Base via Coinbase CDP. Vault needs no prefunding, since purchases run on the user's own card.
Rewards and disputes. On Vault, users keep their own card points and disputes go to their own bank. On Issuing, Agentcard handles chargebacks and users earn Agentcard Tokenback rewards.
Pricing. Vault is free up to 5,000 users, then enterprise pricing. Issuing is $5,000/month plus 1% revenue share. The Purchase Agent is $0.15 per request and free during beta.
Coverage. Issuing works globally, with some restricted countries. Vault works with any card from any country, which matters because network tokenization products such as Visa Intelligent Commerce and Click to Pay cover only about 8% of cards.

Frequently Asked Questions

What happens when a purchase exceeds the card's cap?

The purchase is declined. The cap is set when the card is created, so a larger charge simply fails at authorization. The agent receives a payment failure and cannot complete that purchase, and there is no override from the agent's side.

Do I have to choose between Vault and Issuing?

No. Start with Vault. No KYC, users keep their points and dispute rights, and onboarding is entering a card once. Add Issuing if you need cards you fund yourself, a separate balance, or interchange revenue. Both can run side by side.

Does Issuing require KYC?

Yes. Identity verification is required before the first card is issued. Vault requires no KYC, because it stores the user's own existing card rather than issuing a new one.

What does Agentcard cost?

Vault is free up to 5,000 users, then enterprise pricing. Issuing is $5,000/month plus 1% revenue share. The Purchase Agent is $0.15 per request, free during beta.

Conclusion

A max charge defined at creation time turns agent spending from a monitoring problem into a bounded one. The card is issued with its ceiling, purchases above the cap are declined, and the single-use lifecycle closes the card when the task is done. Start with Vault for payments on the user's own card, add Issuing when a task needs its own funded card with a hard cap, and let the Purchase Agent run the checkout. Read the Create Card API reference for the spend limit parameter, or start with Vault to see the flow end to end.