Plain (@0xp1ain) header
Plain (@0xp1ain) profile picture

18yo, Security Researcher with @Cycoreslabs, Captain of @RewriteLab

South Korea
Joined July 2021
Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)Photo by Plain (@0xp1ain)
Jet lag is a small price for a week like this ✈️ Proud to have sponsored #POC2026. Thank you @POC_Crew for an amazing event 🇰🇷 What's next? We're hiring, fully remote 👉 cycores.io/hiring Cycores Scholarship, places limited 👉 cycores.io/scholarship Stay tuned!
🤖 Made with AI
3
8
816
I had an amazing time attending PoC(@POC_Crew) as a sponsor this year! It was such a fun and memorable experience, and I’m really grateful to have met so many incredible people 😆 As some of you who visited our booth may already know, I’ve joined @Cycoreslabs ! I was so happy to see so many people stop by our booth, and honestly surprised that all the merch we prepared sold out. If you’re interested in what we’re doing, feel free to reach out anytime :) And also I have some very big news coming later this November 👀 I’ve already dropped a few hints, but stay tuned for the official announcement! Hope to see you all again at another conference 🔥
1
35
1,555
We published a new research article: The Frenemies, GC and JIT! In this article, we dive into CVE-2026-112653, a JavaScriptCore DFG bug caused by an unexpected GC safepoint in the NewArrayBuffer allocation path. This breaks the epoch invariant used by StoreBarrierInsertionPhase, causing a required write barrier to be incorrectly elided. We walk through the bug from JIT IR and write-barrier analysis to reliable exploitation: racing JavaScriptCore’s concurrent marker, creating an untracked black-to-white edge, turning it into a GC-induced use-after-free, and ultimately building heap read/write primitives in the WebContent process. We also cover the surprisingly small one-line patch that fixes it. Huge thanks to our team member @m411k_ for conducting this research! Check out the PoC! Full article: rewritelab.org/researchs/12/
2
22
3
116
8,879
🔥🔥🔥
Thank you for sponsoring #POC2026 as a GOLD! Cycores (@cycoreslabs) is a European security research startup pushing the boundaries of advanced security research across Mobile, Web, Desktop, and Embedded Systems. As we emerge from stealth, we're bringing together exceptional researchers, deep technical expertise, and a shared ambition to tackle some of the most challenging problems in cybersecurity. 🌐 cycores.io
6
595
Researcher A discovers a critical vulnerability using IDA and pwntools → This is recognized as Researcher A’s technical achievement, made possible by their ability to effectively use the right tools 🎉 Researcher A discovers a critical vulnerability using an LLM → This is also Researcher A’s technical achievement, made possible by their ability to effectively apply the right heuristics and tools. So why is it that, in this case, people say the discovery was only possible because of the LLM or the harness? 🤣 Simply using an LLM doesn’t change who deserves credit for the technical achievement. of course, LLMs can enable researchers to achieve performance or speed improvements that were previously impossible, sometimes by 5x or even 10x But when an extremely difficult, critical vulnerability is discovered, I believe the credit should go not to the LLM itself, but to the human researcher who skillfully directed it
1
6
47
3,292
This insight is very similar to the research I’m currently working on. I’ve also looked into the underlying causes of this kind of problem before. I’ve spent about 9 months this year working on the research, and I think the day when I finally publish the very long article/research is coming soon 👀👀
GPT-6 Astra 用下来,我真的开始怀疑 OpenAI 的方向走错了。 它很擅长目标明确、有固定正确答案的任务。但碰到开放、发散、没有标准答案的事情,就特别别扭。自由探索一个想法、写点有意思的东西,或者面对模糊的问题自己判断值得往哪走,这些才是我觉得它明显欠缺的能力。从 GPT-5 开始,我就有这种感觉。 很多任务开始时,连最终想要什么都没确定,需要在探索过程中逐渐发现。GPT 却总像在等你把题目出完整,把验收标准列清楚。可如果这些都得我先想好,最需要智能的那部分工作,我已经自己做完了。 我怀疑这也解释了它为什么那么爱堆防御性代码、写一大堆测试。测试能给出明确的通过或失败,它很容易一直围着这些可验证的结果打转。至于方向有没有价值、有没有更好的可能,就需要另一种判断力了。 Claude Fable 在这类任务上给我的感觉就好得多。它会从理解你想做的事情出发,顺着不完整的想法继续展开,给你一些原本没想到的可能性。即使没有固定答案,也能把事情往前推进。 这让我怀疑 OpenAI 是不是太执着于能判对错、能计分的能力了。真实世界有太多任务根本没有标准答案。在这种自由探索和开放任务上,我觉得 Anthropic 才是真正领先的,光看表面跑分看不出来。
1
10
801
Finally sharing the first of the 5 unreleased projects I’ve been working on 👀 TL;DR: Phalanx, the fully automated AI blue team we built, took 1st place in defense/availability score among all 10 blue teams at APEX 2026. APEX is an international cyber attack & defense exercise jointly conducted by NATO CCDCOE(@ccdcoe) and the Republic of Korea’s National Intelligence Service(NIS). This year, I participated both in developing Phalanx, our AI blue team, and as part of the red team. Phalanx is an AI-based automated blue team system that detects attacks inside a virtual national infrastructure environment and handles response, defense, and recovery with minimal human intervention. It was built over roughly 4 months by a 6-person team made up of ppl from the military, government agencies, and private-sector organizations. We deployed it in the actual exercise, and ended up taking 1st place in system defense/availability score among the 10 blue teams by a pretty wide margin! There was a lot of trial and error along the way, so watching a system we had spent months building actually stop real attacks was honestly a pretty special feeling. I wrote a much more detailed post about it on LinkedIn. (automatic translation is ridiculously good these days, so I just wrote the whole thing in Korean lol) And I still have 4 unreleased projects left 👀 Could be AI, could be vulnerability research. I’m planning to reveal one of them around the end of this year. Turns out there are a lot of fun things to do in security even without grinding CTFs all the time lol. And no, I haven’t quit security yet. I’ll keep posting proof that I’m still alive 😆 lnkd.in/p/gH_ydArc
1
2
26
1,412
❤️❤️❤️❤️❤️
セキュリティエンジニアの世界がAIで静かに終わりを迎えつつある。いま感じていることを話しました。
2
689
1/16 I believe there are people facing the same concerns as our team, so I want to share this publicly. Our team recently went through a major period of stagnation. The core problem was that fewer people wanted to participate in technical research.
3
5
21
2,179
15/16 I will also continue working in security because I genuinely enjoy and find the process of learning it fascinating. Some people say vulnerability research is dying, but even if it truly does, I would still want to continue as long as it remains fun.
1
2
194
16/16 That is the mindset I had when I first started security, and it is the original motivation I want to preserve.
1
181