Security policy¶
Reporting a vulnerability¶
We prefer transparency: if you find a security problem in yarlpattern, you are
welcome to open a pull request (or a public issue if a PR is not
practical) so the fix and any discussion can happen in the open.
If you need private coordination first—for example, you believe disclosure should wait until a patch is ready—use GitHub’s security advisory flow:
https://github.com/chad-loder/yarlpattern/security/advisories/new
In either case, we aim to acknowledge reports within 72 hours.
Supported versions¶
yarlpattern is pre-1.0, so only the latest published minor version is
actively maintained. Once a stable 1.x line ships, this policy will be
updated to cover at least the current and previous minor releases.
Scope¶
In scope:
- Regular-expression denial-of-service (ReDoS) from a benign-looking pattern that compiles to a catastrophic-backtracking regex.
- Spec deviations that could let an attacker bypass a routing check —
e.g. a URL that should not match a pattern but does, or vice versa,
when the application uses
URLPatternfor authorization decisions. - Memory exhaustion from inputs accepted by the constructor.
Out of scope:
- Issues in dependencies (report to
yarlorregexdirectly). - ReDoS that requires the application to compile patterns supplied by an untrusted user. Treat user-supplied patterns the same way you would treat user-supplied regular expressions: don't.
- Self-DoS from passing absurd inputs to the API.