mime-mail is a Haskell library used to compose and render emails. It is typically used in server-side software to send automated emails or construct emails from user supplied inputs.
While working on a website where I was constructing emails with POST data supplied by the user, I discovered that the mime-mail package was vulnerable to a header injection vulnerability. When browsing mime-mail's source code, I noticed that email headers were not being sanitized before being rendered. This meant that an attacker could insert additional email headers by injecting CLRF (
\r\n) characters between the malicious headers. As a result, attackers could control parameters including the From, To, Reply-To, CC, BCC, and Subject fields. Since the emails are sent from the website, attackers could use this vulnerability to send spam, phish users for credentials, leak private information, or deface the site.
We can demonstrate the vulnerability with ghci, Haskell's REPL: