RAT is a development tool.  As such we expect our users to properly
sanitize any XML and XSLT input.

On another project I work on Security recently suggested they add a
statement in the SECURITY.md file to explicitly call out this expectation.

I think we should do the same.  I believe that our text should read
something like.


   - Configuration files and XSLT documents passed to RAT are
operator-controlled configuration, not request input. Reports claiming SSRF
/ path traversal via these resolvers,  on the assumption that the resource
name is attacker-controlled, are out of scope under the documented threat
model: xml and xslt authorship and resource configuration are privileged
operations.
   - Applications that thread untrusted input into XML configuration or
XSLT documents should validate that input before passing it to RAT; the
responsibility for that validation rests with the application, not with RAT.


I think that if we add a security file with the above statments we can
disable the XXE_DOCUMENT issue in spotbugs.

Claude

-- 
LinkedIn: http://www.linkedin.com/in/claudewarren

Reply via email to