How to differentiate refresh_token from access_token at validation stage in security server?

Question

I know the difference in concept and purpose, but what happen at syntax level?

Let's suppose a login endpoint with basic tokens response:

http://acme.com/login

{
    "access_token": "AYjcyMzY3ZDhiNmJkNTY",
    "refresh_token": "RjY2NjM5NzA2OWJjuE7c",
    "expires_in": 3600
}

Both tokens are been generated by security server using using its algorithms (secret and claims)

So when access_token expires, I need to renew it using some endpoint like

http://acme.com/oauth/token

Sending the refresh_token

---

At this point born my question

Assuming that these tokens have different string values and different expiration times, but they have been generated with the same secret, claims or algorithms

And I need to implement a refresh token generation and validation at security server backend.

How to differentiate access_token from refresh_token ?

---

Current Workaround

Add an extra claim (e.g type=refresh) just for refresh_token, so when I receive a token string in the security server and I decode it, this claim is useful to determinate if the received string is a refresh_token and not an access_token.