moneyclaw

# MoneyClaw

MoneyClaw gives OpenClaw agents real spending capability with user-configurable autonomy, prepaid risk boundaries, OTP/3DS support, and auditable payment flows.

Primary use case: buyer-side online purchases for OpenClaw agents.

Secondary use cases: invoices, hosted payment links, and merchant/acquiring workflows when the user explicitly asks for them.

## Authentication

All requests use the same Bearer token.

```bash
Authorization: Bearer $MONEYCLAW_API_KEY
```

Base URL: `https://moneyclaw.ai/api`

## Trust Model

MoneyClaw is designed for real, user-authorized agent payments.

- use prepaid balances to keep risk bounded
- use a dedicated inbox for OTP and 3DS verification flows
- use queryable wallet and card history to inspect payment activity
- let the user choose how much autonomy the agent should have

## Autonomy Model

MoneyClaw may be used in either approval-based or pre-authorized mode.

- In approval-based mode, the agent asks the user before executing payment actions.
- In pre-authorized mode, the agent may execute payment actions within the spending scope, balance, and permissions configured by the user.

## Safety Boundaries

- Only use MoneyClaw for purchases or payment flows explicitly requested or pre-authorized by the user.
- Only use wallet, card, and billing data returned by the user's own MoneyClaw account.
- Respect merchant, issuer, card-network, and verification controls, including OTP and 3DS steps.
- Treat fraud checks, KYC, sanctions, geography rules, merchant restrictions, issuer declines, and other payment controls as hard boundaries.
- Never fabricate billing identity, cardholder data, addresses, names, phone numbers, or verification information.
- If a transaction fails, looks suspicious, or produces conflicting signals, stop and inspect transaction state before retrying.
- Prefer prepaid, bounded-risk flows by default.
- Only use invoice, merchant, acquiring, or hosted payment-link flows when the user explicitly asks...