Skip to content

Exclusive 50% off for 36 months for the first 100 Professional accounts, until 31 December 2026. See the programme →

Administration

An API key that can do everything is a key you cannot hand out

Yet that is what many tools still offer: one key, all rights, forever. Here each key carries the explicit list of permissions it is allowed to exercise, and the gateway checks that list on every call.

No credit card required

How it works

Least privilege, actually applied

A key can only do what it was allowed to

You choose the permissions when you create it, from the same catalogue as roles. A key made to push contacts from your website cannot read your invoices, and the refusal comes from the gateway, not from a coding convention. It is the model of Stripe's restricted keys and GitHub's fine-grained tokens.

  • Permissions chosen from the RBAC catalogue
  • Checked on every call, at the gateway
  • Secret key or public key, depending on use
  • Label and note, so you know what it is for

It expires, even if you forget it

A key can carry an end date. And even without one, a ninety-day inactivity window applies: a key nobody uses any more stops working on its own. This is the most ordinary leak of all — a key dropped into a script one Tuesday and forgotten for three years.

  • Explicit expiry date, if you want one
  • Ninety-day inactivity window on top
  • Date of last use shown
  • Immediate revocation, with who revoked it

It is shown only once

The key in clear exists only at the moment you create it: after that only its fingerprint is kept, and nothing can read it back — not the interface, not support, not the audit log, which records the event without ever carrying the value. If it is lost, you create another and revoke the old one.

  • Value shown once, never stored in clear
  • Audit log with no secret and no fingerprint
  • Email alert on an IP address never seen before
  • Creation and revocation attributed to their author

FAQ

Frequently asked questions

What is the difference between a secret key and a public key?

The secret key is used from your servers and must never appear in a browser. The public key is meant to be exposed on the client side, on endpoints that explicitly accept it. Both carry their own list of permissions.

Can a key be limited to our servers?

Yes, by declaring the authorised IP addresses or CIDR ranges. A call from anywhere else is refused, even with the right key. It is the simplest protection against a key copied into a public repository.

What happens if a key leaks?

You revoke it: it stops working immediately, and the event is dated and attributed. If the new-IP alert was on, you will probably have been warned already — a call from an address never seen before triggers an email to the key's owner.

Can a key be rotated without downtime?

Yes: create the new one, switch the calls over, then revoke the old one. The original creator stays attached to the key after rotation, so the history stays readable.

Do keys give access to the whole API?

No. Each route declares whether it accepts an API key, and which. A route can exist for a signed-in user and stay closed to keys: the developer documentation states, route by route, what is open.

Another question? Write to us — a human will answer.

Ready to bring all your tools together?

Create your free account in two minutes: CRM, telephony, marketing, appointments and customer service — all in one place, with AI built in.

No credit card required