Flag Proxy
WEB | 471 pts - 41 solves
Last updated
WEB | 471 pts - 41 solves
Last updated
Description: I just added authentication to my flag service (server-back) thanks to a proxy (server-front), but a friend said it's useless... Site: http://flag-proxy.challs.olicyber.it Author: @Giotino
Looking at the challenge, we see that the website it's powered by Express.js. Looking in the source code we see that the endpoints it accepts are: /flag
and /add-token
. To get the flag we have to make a request with the ?token
parameter which value will then be passed to the Authorization
header that will be sent to the backend; which to be valid must be inside the tokens[]
array. We can add values to tokens[]
array by making a request at /add-token
with the parameters: ?token
(the token we want to set) and ?auth
(some sort of password). Looking at the logic of these two endpoints, it would seem almost impossible to find an attack vector. But looking beyond the index.js's we notice that in the server-front, http-client.js
manages and parse all the requests and responses of the application. Looking more carefully we notice how on line 55-58 it checks that in the headers we pass, there is no LF character to protect from http request smuggling attacks. However, the program only checks for the sequence , but not the character alone which probably being HTTP 1.0 and an old version of node.js, could work.
Also we need to find a way to chain requests in HTTP 1.0 (modern http-smuggling techniques won't work, i.e: Transfer-Encoding: chunked
), this can be done by setting the header Connection: keep-alive
.
Let's try to craft a double smuggling request with docker container running in local to analyze logs:
It's working! Now, lets make a request to /add-token
with an arbitrary token, to be able to write our token inside tokens[]
array, bypassing app checks (AUTH param), and later getting the flag:
flag{sanity_check}