JWT labs

A practical playbook — from recon to proof of concept.

1. 1

   Decode the token

   Split the three segments and base64url-decode the header and payload. Check the algorithm, the kid field and claims like role or user_id.

   echo $JWT | cut -d. -f1 | base64 -d   →   {"alg":"RS256","kid":"1"}

2. 2

   Try alg=none

   Change the algorithm to none, edit the payload (e.g. role=admin) and send the token with no signature. Vulnerable libraries accept it.

   header {"alg":"none","typ":"JWT"}  ·  payload {"user":"admin"}  ·  empty signature: header.payload.

3. 3

   Algorithm confusion RS256→HS256

   If you know the RSA public key, sign an HS256 token using it as the HMAC secret. A server that verifies with the public key deems it valid.

   jwt_tool $JWT -X k -pk public.pem   (confusion attack)

4. 4

   Crack weak HS256 secrets

   Dump the JWT to hashcat and try a wordlist. A cracked secret lets you sign any token.

   hashcat -a 0 -m 16500 jwt.txt rockyou.txt

5. 5

   Inject kid/jku

   Tamper kid to point at a known key file (/dev/null → empty secret) or jku/x5u to a JWKS you host.

   {"alg":"HS256","kid":"../../../../dev/null"}  →  sign with an empty secret