Two layers decide what a request can reach:
- Account capabilities are what your account as a whole may do.
- Credential scopes are what one OAuth client may do inside it. They are validated against your capabilities when the credential is created.
effective access = credential scopes ∩ account capabilitiesThe vocabulary
| Value | Covers |
|---|---|
payouts | Money movement: payouts, batches, beneficiaries, instruments, balances, accounts and funding, FX, pricing, corridors and the sandbox seeders. |
intelligence | Anton Intelligence: risk evaluations and verifications. See the Intelligence overview. |
Compliance-only accounts
An account can hold only intelligence. It buys screening and evaluation without moving money through Anton: every payout-surface endpoint returns 403 capability_required, so the account cannot create payouts, beneficiaries, instruments, batches or FX exchanges.
Webhooks and reference data (currencies, countries, payment methods) work on every account. Payout events never fire for an account that cannot create payouts.
A compliance-only integration is the Intelligence quickstart: mint a token, POST /v1/intelligence/evaluations, read evaluations back. To add payouts to a compliance-only account, contact Anton through help.antonpayments.com.
Scopes in tokens
A token carries exactly the scopes its credential was created with, as a space-delimited scope claim (RFC 6749 §3.3), echoed in the token response. POST /oauth/token has no scope request parameter.
{
"access_token": "eyJhbGciOiJFUzI1NiIs...",
"token_type": "DPoP",
"expires_in": 3600,
"scope": "intelligence"
}Give compliance tooling an intelligence credential and your payment service a payouts credential, so a leaked compliance credential cannot move money.
When a call is refused
Both layers fail with 403 permission_error and a stable code.
| Code | Layer | Meaning |
|---|---|---|
capability_required | Account | Your account does not hold the capability this endpoint needs. Typical for a compliance-only account calling a payout endpoint. |
insufficient_scope | Credential | Your account holds the capability, but this credential was created without the matching scope. Use a credential that has it. |
{
"error": {
"type": "permission_error",
"code": "capability_required",
"message": "this endpoint requires the \"payouts\" account capability",
"doc_url": "https://docs.antonpayments.com/api-reference/errors#capability_required",
"request_id": "req_01JD3XB7Q2"
}
}Neither code is retryable. The same request fails identically until the capability or the credential changes. See Errors.
What each surface needs
| Surface | Needs |
|---|---|
| Payouts, batches, beneficiaries, instruments, balances, accounts, FX, pricing, corridors, sandbox seeders | payouts capability and scope |
| Intelligence evaluations | intelligence capability and scope |
| Verifications | intelligence capability and scope |
| Intelligence reviews | intelligence capability and scope, plus the Reviews API and Webhooks add-on. Exports also need the Audit Pack. Without the add-on the call is 403 module_not_enabled. |
| Webhooks, reference data | Any authenticated credential |
What a credential can never do
Two kinds of operation need a signed-in person, whatever the scope:
- Approval decisions. Approving or rejecting a payout needs a human identity, because maker-checker means nothing for a credential.
- Reading personal data back. Beneficiary personal data, verification names and captured documents are dashboard-only, restricted by role and audit-logged on every access. A credential can write personal data (
PATCH /v1/beneficiaries/{id}/pii) but never read it, so a leaked credential cannot harvest identities.
Responding to an RFI is dashboard-only too, and needs rfis.respond (Admin or Operator).