Client credentials can be used for machine-to-machine communication authentication. Clients can authenticate themselves using service-accounts; standard client_id + client_secret is not sufficient. This behavior is due to providers only being able to have a single secret at any given time.
Hence identification is based on service-accounts, and authentication is based on App-password tokens. These objects can be created in a single step using the *Create Service account* function.
An example request can look like this:
```
POST /application/o/token/ HTTP/1.1
Host: authentik.company
Content-Type: application/x-www-form-urlencoded
grant_type=client_credentials&
client_id=application_client_id&
username=my-service-account&
password=my-token
```
This will return a JSON response with an `access_token`, which is a signed JWT token. This token can be sent along requests to other hosts, which can then validate the JWT based on the signing key configured in authentik.
### JWT-authentication
Starting with authentik 2022.4, you can authenticate and get a token using an existing JWT.
(For readability we will refer to the JWT issued by the external issuer/platform as input JWT, and the resulting JWT from authentik as the output JWT)
To configure this, the certificate used to sign the input JWT must be created in authentik. The certificate is enough, a private key is not required. Afterwards, configure the certificate in the OAuth2 provider settings under *Verification certificates*.
With this configure, any JWT issued by the configured certificates can be used to authenticate:
Alternatively, you can set the `client_secret` parameter to the `$inputJWT`, for applications which can set the password from a file but not other parameters.
Input JWTs are checked to be signed by any of the selected *Verification certificates*, and their `exp` attribute must not be now or in the past.
To do additional checks, you can use *[Expression policies](../../policies/expression)*:
