1. people who think these bugs aren't practically exploitable are outright wrong. to be clear, we've exploited php2shell, brotli2shell, java4shell, and ruby2shell against live remote targets, not just labs. haven't tried nginx2shell yet, but i suspect it's exploitable too (should have more to share soon).View quoted post on X
2. the distinction is that exploitation isn't trivial or easily portable. unlike react2shell, you can't just spray a generic poc across vulnerable servers and expect shells. these bugs often require significant target-specific work, and success on one target doesn't guarantee success on another. (prove me wrong)
3. i don't think threat actors can easily exploit these at scale, but spending a day or two on a high-value target to get a shell or leak sensitive data is a realistic possibility.
4. should you patch? absolutely. will someone exploit you? idk, depends on who you are. if you're openai/anthropic/inference-provider/bank/crypto/other-critical software, and if i am threat actor i would come after you, because the economics of targeted exploitation are very different from mass exploitation.
5. are these bugs overhyped? i think so, if we're talking about react2shell level exploitability and mass exploitation potential. if 2shell is just a naming convention, that's fine.
but if the name is meant to imply similar ease of exploitation or widespread impact, i think the criticism is fair, that is what happening here, people would draw that comparison, even though the exploitation characteristics are very different.
Latest posts 5
Nearly 376,000 FortiGate hosts expose a management port that should stay private. A single TLS ClientHello also reveals their firmware generation, helping defenders assess potential exposure to known exploited flaws.View quoted post on X
Read the full research: vulncheck.com/blog/fortigate…
💯 my points.
My two cents after dealing with Java4Shell and Nginx2Root / Nginx2Shell with our customers:View quoted post on X
I think it is more important than ever that security firms and frontier labs be critical on the vulnerabilities being published. Especially if there are hardware constraints, scenario constraints or anything that reduces the chances of exploitability.
At least for the next few months (if not more), we are going to see rise on these kind of vulnerabilities being released. If we keep crying wolf on every “devastating” bug, defenders will stop listening. And when a genuinely exploitable RCE hits something like Apache or React again, vulnerability management teams won’t respond with the urgency it demands.
Security firms and frontier labs need to stay grounded. Every disclosure should come with honest, specific context about exploitability. Our job is to help defenders, not to shout from the mountaintop.
Thank you 💯 Or disable http/2 until you take the time to properly deploy upgrades.
And also: http/2 is disabled by default.
When will we stop cry wolf?
#Stop2shell
And also: http/2 is disabled by default.
When will we stop cry wolf?
#Stop2shell
The RCE for nginx needs 128GB of HTTP/2 traffic to trigger and specific CPU instructions to be exploitableView quoted post on X
I hope I saved you some time.
CVE-2026-123007 was also independently found by Rapid7, Wiz and Xint.io. Great work, all!View quoted post on X