SpecterOps has 3 free on-demand courses in its Tradecraft Academy. Kubernetes for Red Teamers is a solid intro to attacking Kubernetes and Mythic.
I also patched the lab for KVM/libvirt:
https://t.co/X3LwqNziZt
#RedTeam#Kubernetes
Original PoC by Aniq F.
https://t.co/i4aXOfx5CZ
Patched forks if you want to test before the PRs merge.
https://t.co/LzmbOQeulp
https://t.co/URPxCe8yIr
CertiGhost (CVE-2026-54121) was published on 24 July by H0j3n and Aniq F. I expected to spend five minutes running their PoC in my Ludus lab. Every run failed in the same place with STATUS_NOLOGON_SERVER_TRUST_ACCOUNT.
https://t.co/kULeSn224k
With both patched, the CA cleared both callbacks and issued a certificate. A workstation target completed PKINIT and returned its NT hash and ccache. Three PRs followed.
Impacket https://t.co/p5h1gUe57U
CertiGhost fix https://t.co/SI4EIbjG5G
Debug https://t.co/3RyMZULoKR
I bypassed Skillable's per-user lab limit by changing a userId the browser hands over, then used a colleague's allocation with his consent. Skillable says it is SCORM's fault, not theirs, and will not fix. So I disclosed it. CVE-2026-56877.
https://t.co/Bjb6FwIjR4
A real root-level bug chain, acknowledged by the maintainer and the deployment operator. Still no CVE.
Declined on scope, rejected twice as “appears to be a test”, now parked at MITRE. Fixed and merged upstream; the ID field still blank.
https://t.co/t3bj9kst3m
#cve#CRTO
Passed the CRTO. Five years late, but passed.
Retrospective on what changed between 2020 and 2026: platform, tradecraft, exam. Plus two security issues reported under coordinated disclosure.
https://t.co/HAB6aVuVea
Same appliance tested twice at two different orgs.
First time: 20+ findings from VA scans and a ChatGPT prompt. Second time: 5 CVEs in 5 days, vendor patches, halted go-live.
I wrote up what changed.
https://t.co/7gg68iMsDn
2/2 Grok cited 119 sources with CLI flags that don't exist. Claude caught the hallucination… then suggested its own unverified workarounds. The actual fix? Plain old human debugging instinct.
Full write-up: https://t.co/j6LaJV2yvI.
1/2 War diary from LLM-assisted pentesting: needed a vulnerable che-machine-exec instance for my Metasploit PR (CVE-2025-12548, unauth RCE in Eclipse Che). Asked Grok 4 and Claude Sonnet 4.6. Both confidently wrong.
I think lots of developers find it surprising that resource limits apply across setuid/setgid/filecap exec boundaries. Here is the quick and dirty tool I use to find interesting errors to investigate. https://t.co/xuJwh9rcPS