Minitar Security Policy
Minitar aims to be secure by default for the data inside of a tar file.
LLM-Generated Security Report Policy
Absolutely no security reports will be accepted that have been generated solely by LLM agents. There must be a human that confirms the issue.
Supported Versions
Security reports are accepted only for the most recent major release.
Reporting a Vulnerability
Report vulnerabilities via the Tidelift security contact. Tidelift will coordinate the fix and disclosure.
Alternatively, create a private vulnerability report with GitHub.
Exclusions
There are several classes of potential security issues that will not be accepted for Minitar, because any issues arising from these are a matter of the library being used incorrectly.
Minitar does not perform validation or sanitization of path names provided to the convenience classes Minitar::Output and Minitar::Input, which use Kernel.open for their underlying implementations when not given an IO-like object.
Improper use of these convenience classes with arbitrary input filenames may leave your your software to the same class of vulnerability as reported for Net::FTP (CVE-2017-17405). If the input filename argument starts with the pipe character (|), the command following the pipe character is executed.
Additionally, the use of the open-uri library (which extends Kernel.open with transparent implementations for Net::HTTP, Net::HTTPS, and Net::FTP), there are other possible vulnerabilities when accepting arbitrary input, as detailed by Egor Homakov.
These security vulnerabilities may be avoided, even with the Minitar::Output and Minitar::Input convenience classes, by providing IO-like objects instead of pathname-like objects as the source or destination of these classes.