<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Samson Chen on Medium]]></title>
        <description><![CDATA[Stories by Samson Chen on Medium]]></description>
        <link>https://medium.com/@samson_63611?source=rss-6f5c7211d8a0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*0jTnWhQZWlNGY6T0</url>
            <title>Stories by Samson Chen on Medium</title>
            <link>https://medium.com/@samson_63611?source=rss-6f5c7211d8a0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 10:19:28 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@samson_63611/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[I Got Phished: A Senior Engineer’s AMOS Incident Report]]></title>
            <link>https://medium.com/@samson_63611/i-got-phished-a-senior-engineers-amos-incident-report-b54f667c9789?source=rss-6f5c7211d8a0------2</link>
            <guid isPermaLink="false">https://medium.com/p/b54f667c9789</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[malware]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[phishing]]></category>
            <category><![CDATA[macos]]></category>
            <dc:creator><![CDATA[Samson Chen]]></dc:creator>
            <pubDate>Sat, 19 Sep 2026 08:03:32 GMT</pubDate>
            <atom:updated>2026-09-20T08:15:11.808Z</atom:updated>
            <content:encoded><![CDATA[<h3>How Mac Users Get Phished: What Happened to Me</h3><h3>From clicking Download Now to deciding to wipe the machine, it took 5 minutes</h3><p>*This article was originally published in Traditional Chinese. You can read the <a href="https://medium.com/p/39f79565d10d">original version here</a>.*</p><p>(I wrote this article in Traditional Chinese. The English text was translated and proofread with AI. The content is mine.)</p><p>I have over 20 years of experience as a software developer, and I still got phished. I have never written anything on Medium before, but I decided to write this embarrassing episode down so that fewer people end up in the same place. This incident broke a few assumptions I had been carrying around, all of which are clearly wrong today:</p><ol><li>(mildly wrong) Attackers don’t particularly target mac users</li><li>(moderately wrong) macOS is a reasonably trustworthy environment</li><li>(badly wrong) Google keeps the bad sites away from you</li></ol><p>The technique that got me has a name: AMOS, Atomic macOS Stealer. The name has macOS in it, so it is obviously aimed at mac users. Supposedly (I haven’t verified this) it is called Atomic because the original main target was the Atomic Wallet crypto wallet.</p><p>Put simply, AMOS uses social engineering to get you to hand your own data over. The whole thing happens very fast. In the few seconds it took me to realize something was wrong, my data was already on its way out. Within 5 minutes I decided to pull the network, shut down, wipe and reinstall macOS. There was no actual loss in the end, but I then spent an entire day changing or resetting everything that could have leaked. More on that below.</p><h3>How it started</h3><p>It started with an article by a user on medium.com:</p><p><a href="https://liu-111.medium.com/the-best-tuis-in-2026-why-developers-are-quietly-abandoning-their-desktop-apps-5acf3ddaec97">The Best TUIs in 2026: Why Developers Are Quietly Abandoning Their Desktop Apps</a></p><p>The article is a straightforward introduction to various Text User Interface tools, things like LazyGit and btop. The fourth one he covers is called lazydocker. I didn’t do anything about it right away, and I didn’t follow the links in his article either. A day later I suddenly remembered it and wanted to try it out. The name is easy enough to remember, so I figured I would just Google for the source. I searched for lazydocker:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-7Bzch82wKQ4wqrK0pf8kw.png" /></figure><p>The first result in my list was a site called lazydocker[.]com (that’s the hook). The second one is actually the real one, but look at the descriptions: the second one reads vague and unclear on Google, and the first one reads perfectly clear.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*unxW4A_OtpaJBUIy-IbKDw.png" /></figure><p>Looks legitimate enough, doesn’t it? So naturally I saw that Download Now and clicked it:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rK6ekJce6tNxmz47TCd41Q.png" /></figure><p>At this point I already felt something was off, because the installation method was a shell command. I was suspicious, but plenty of real tools install this way too, Homebrew, AWS CLI, Miniconda. So it didn’t quite register. I pasted the command in, and the OS popped up asking for my machine password (the OS, not something else). I didn’t think much of that either, because the instructions on the page had said it would ask for a password. (What a textbook case of falling for my own habitual thinking.) Then I read the text properly and it said Chrome Safebox, and that is when it hit me that I had been phished. The whole thing was only a few seconds, and that data should already have been sent out. Right then I decided to go all the way: pull the network, shut down, wipe and reinstall macOS.</p><h3>Looking back</h3><p>Why could this fake site stay up for so long without being flagged as malicious by Google, <a href="https://urlscan.io/">urlscan.io</a> or <a href="https://www.virustotal.com/gui/home/upload">VirusTotal</a>? According to <a href="https://github.com/jesseduffield/lazydocker/issues/666">issue #666</a> in lazydocker’s official GitHub repo, it showed up August 2025, and phishing reports appeared as early as February 2026. It uses a few methods to stay alive:</p><ol><li>That Download Now button doesn’t always lead to the phishing install page. Most of the time it redirects to a CPC ad page. Which means there’s a good chance it only ever gets labeled as a content farm chasing traffic and ad clicks, rather than a malicious site.</li><li>It hides behind CloudFlare, and uses CloudFlare’s WAF to block sources that might scan it by treating them as “attacks”. So the systems that would spot its malicious behavior get classified as attackers instead.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bLwTC6C5r8P98oxDFBIwHw.png" /></figure><p>Above is VirusTotal’s analysis of the final URL that actually delivers the phishing payload. Everything reads Clean. But the thing to look at is the HTTP Status 520 in the top right corner: that request was blocked by the WAF. That is exactly why it can stay up without being discovered. This kind of phishing is clearly aimed at engineers like us, the overconfident ones who think we know better.</p><h3>Gemini vs Claude</h3><p>As a side note, I fed this whole analysis to both Gemini and Claude. Both could tell it was phishing, but Claude’s analysis was considerably better than Gemini’s. Gemini was conservative to the point of giving advice on the assumption that every piece of data had gone out. Claude, by contrast, did a good job of working out which risks were high and which were low, and it also worked out that the malicious site was hiding behind CloudFlare and told me to report it to CloudFlare. That part mattered, because once I reported it to CloudFlare directly, the site was suspended within 6 hours.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zREfO-YOgMwh6kmuNf6_Lw.png" /></figure><p>As for Google, I reported it to Google Safe Browsing several times and had friends report it too. A full 24 hours went by with no response and it was still showing up in search.</p><h3>What got stolen</h3><p>The malicious code was obfuscated in multiple layers, so it’s hard to analyze what it actually sent out, and I’m not a dedicated lab that can set up a sandbox to observe it. But my assessment is that the following went out:</p><ol><li>All Web Session Tokens</li><li>The contents of .env/.ssh/.aws</li><li>VPN scripts, including VPN endpoints</li><li>keychain-db, including old password data (if you haven’t moved to iCloud), Certificates (Apple Developer ones especially), and SMB service passwords</li></ol><h3>Cleaning up afterwards</h3><p>It happened too fast, so the only option was to assume the worst. Here is my cleanup list:</p><ol><li>Wiped and reinstalled macOS</li><li>Changed passwords on every cloud service that still had a live Web Session, and revoked all macOS logins</li><li>Revoked and rotated all API keys, the AI service ones especially</li><li>Revoked the cloud-side permissions of every key in .ssh, and moved everything to Secretive</li><li>Revoked all AWS-issued Access Credentials</li><li>Revoked and replaced all VPN certificates</li><li>Revoked all Apple Developer certificates</li><li>Changed all SMB service passwords</li></ol><h3>What was safe</h3><p>The following I judged to be safe:</p><ol><li>Sessions on other devices for the same cloud services (though some cloud services revoke every session once you change the password)</li><li>Passwords and TOTP keys stored in a password manager with a Master Passphrase. I use Zoho Vault and iCloud Passwords. Unlike Chrome Safebox, these two can’t be scooped up in one go, so they are a lot safer.</li><li>Cloud services whose Web Sessions had already expired.</li><li>Services logged into via SSO (AWS Identity Center, for example)</li></ol><h3>My recommendations</h3><p>After going through this, here is my honest advice.</p><ol><li>When installing anything, if it installs via shell, be careful, then be more careful. If it can be installed through a channel like Homebrew, use that. Whenever you see something unfamiliar, ask first whether there’s a brew install method. Avoid the shell install where you can.</li><li>Never let your browser manage your passwords, Chrome especially. Chrome’s Safebox is very easy to scoop up wholesale. Safari uses iCloud, so it is relatively safer. Better still, use a dedicated password manager such as <a href="https://www.zoho.com/vault/">Zoho Vault</a> or <a href="https://1password.com/">1Password</a>, the kind with a Master Passphrase or TouchID/FaceID in front of it. I use Zoho Vault myself, which is why this didn’t turn into a password incident.</li><li>If a cloud service offers passkeys, use them. If it has MFA, turn it on. These leaks move fast when they happen, and those extra layers save you a lot of trouble.</li><li>For SSH keys, use something like <a href="https://github.com/maxgoedjen/secretive">Secretive</a> that works as an agent, rather than keeping a private key on the machine itself.</li><li>Consider installing a firewall like <a href="https://objective-see.org/products/lulu.html">LuLu</a>. If something does start sending data out, it can stop it at the first moment. Though once phishing has been invited in, you still have to reinstall the OS.</li><li>Have a good backup setup. Good enough that you can decide to wipe and reinstall immediately without losing data. In my case everything syncs and backs up to a NAS.</li></ol><h3>A hostile environment</h3><p>Think about it for a moment. If a malicious attacker deliberately writes a small tool, wraps it in layer after layer of packaging and various methods of blocking scanners, keeps the tool itself harmless so it looks like an innocent third party, writes introductory articles about it all over the place, and then uses indirect means to lead you down the wrong path yourself, you could very easily end up where I did. Beyond your own alertness, the best protection comes from everyday habits. Don’t rely on a single layer of protection, and stay alert. That’s my advice to you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b54f667c9789" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[我被釣魚了：一個資深工程師的 AMOS 中招紀錄]]></title>
            <link>https://medium.com/@samson_63611/%E6%88%91%E8%A2%AB%E9%87%A3%E9%AD%9A%E4%BA%86-%E4%B8%80%E5%80%8B%E8%B3%87%E6%B7%B1%E5%B7%A5%E7%A8%8B%E5%B8%AB%E7%9A%84-amos-%E4%B8%AD%E6%8B%9B%E7%B4%80%E9%8C%84-39f79565d10d?source=rss-6f5c7211d8a0------2</link>
            <guid isPermaLink="false">https://medium.com/p/39f79565d10d</guid>
            <category><![CDATA[information-security]]></category>
            <category><![CDATA[phishing]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[macos]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Samson Chen]]></dc:creator>
            <pubDate>Sat, 19 Sep 2026 08:03:00 GMT</pubDate>
            <atom:updated>2026-09-20T08:13:16.330Z</atom:updated>
            <content:encoded><![CDATA[<h3>Mac 使用者是怎麼被駭客釣魚的：我的親身經歷</h3><h3>從點下 Download Now 到決定重灌，只有 5 分鐘</h3><p>*本文另有英文版本：<a href="https://medium.com/p/b54f667c9789">English version</a>*</p><p>(本文僅使用 AI 校稿，全文由本人親自寫出，無 AI 生成內容。)</p><p>身為一個超過 20 年經驗的資深軟體系統開發者，居然被駭客釣魚成功，為了不要有更多的受害者出現，從來沒在 Medium 上面寫過東西的我，決定把這個丟臉的事蹟寫給大家，希望不要有更多人受害。這一次事件，打破我過去很多顯然在今天絕對是不正確的認知：</p><ol><li>(小錯) 駭客不太喜歡針對 mac 的使用者進行攻擊</li><li>(中錯) macOS 環境的安全性比較可以信賴</li><li>(大錯) Google 會幫大家把壞網站擋在前面</li></ol><p>我中招的釣魚手法有個專有名詞叫 AMOS，Atomic macOS Stealer。專有名詞裡就帶個 MacOS 顯然就是衝著 mac 的 user 來的。據說 (我沒有查證) 會叫 Atomic 的原因是因為一開始的主要目標是 Atomic Wallet 的加密貨幣錢包。</p><p>簡單的來說, AMOS 是透過社交行為讓你自己把東西送上門，從頭到尾發生的速度非常的快，等到我感覺不對的幾秒鐘內，我的資料已經開始往外送，最後我在 5 分鐘內決定斷網、關機、重灌整套 macos。雖然沒有實質損失產生，但後面整整花了一整天去把我所有可能洩漏的東西給變更或重置，文後會詳述。</p><h3>怎麼開始的</h3><p>源頭來自於 medium.com 上一位 User 寫了一篇文章:</p><p><a href="https://liu-111.medium.com/the-best-tuis-in-2026-why-developers-are-quietly-abandoning-their-desktop-apps-5acf3ddaec97">The Best TUIs in 2026: Why Developers Are Quietly Abandoning Their Desktop Apps</a></p><p>這篇文章也就是一個非常單純的介紹各種 Text User Interface 的各種工具例如 LazyGit 或 btop 這種東西，然而他介紹的第四樣叫 lazydocker 我看完後沒有馬上做什麼，也沒有順著他文章裡的連結往下走。隔了一天以後，突然想起來想要試一下這個東西，因為名字並不難記，所以我想用 Google 去找軟體來源就好，所以就直接用 lazydocker 去 Google 上找:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-7Bzch82wKQ4wqrK0pf8kw.png" /></figure><p>在我的搜尋清單裡第一個是一個 lazydocker[.]com 的網站 (這個就是釣魚勾)，其實第二個才是真的，但如果看簡介，第二個在 Google 裡就是寫的不清不楚，第一個清清楚楚。</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*unxW4A_OtpaJBUIy-IbKDw.png" /></figure><p>點進去後很像一回事吧! 然後想當然爾我就看到那個 Download Now 就點下去：</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rK6ekJce6tNxmz47TCd41Q.png" /></figure><p>這時我已經覺得怪怪的，因為安裝的方式用的是 Shell，我當時雖然懷疑，但因為有一些工具程式例如 Homebrew, AWS CLI 或是 Miniconda 也是用 Shell 方式安裝，所以我當下沒太反應過來。等我把 Command 貼進去，OS 就跳出來問我電腦密碼 (是 OS 不是別人)，我當下也沒想太多，因為上面的說明就說會問密碼 (唉! 真是中了心理學上的習慣性思維的招呀)，等到我看清楚字寫著 Chrome Safebox，我開始驚覺糟糕，被釣魚了，這中間只有幾秒鐘的時間，那些資料就應該被送出去了。我當下立馬決定斷網、關機、重灌這麼大的動作。</p><h3>事後諸葛</h3><p>這個假網站為何可以存在那麼久而不被 Google, <a href="https://urlscan.io/">urlscan.io</a>，<a href="https://www.virustotal.com/gui/home/upload">VirusTotal</a> 標記為惡意網站呢 ? 按照 lazydocker 官方 GitHub Repo 的 <a href="https://github.com/jesseduffield/lazydocker/issues/666">issue #666</a>，2025年八月就出現了，而釣魚的報告在2026年二月就出現過。它用了幾種方法讓這個釣魚站台可以一直躲過被揪出：</p><ol><li>Download Now 那個按鈕其實不一定會到釣魚安裝的頁面，大部份的情況是轉到 CPC 廣告的網頁，也就是說，它很大的機會只會被標為內容農場賺流量與點擊廣告的那一種，而不是惡意網站。</li><li>它躲在 CloudFlare 後面，然後用了 CloudFlare 的 WAF 把可能會掃描它的來源給當成「攻擊」擋掉，也就是說會看出它有惡意的系統被它當成攻擊。</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bLwTC6C5r8P98oxDFBIwHw.png" /></figure><p>上面這個是 VirusTotal 對最後那段真正植入 Phishing 的 URL 做解析，上面寫的全部是 Clean，然而重點在右上角那個 HTTP Status 520，這個 request 被 WAF 給擋掉了。這也就是為何它可以一直存在而不被發現。這種釣魚法顯然就是衝著我們這種過度自信又自以為是的工程師來的。</p><h3>Gemini vs Claude</h3><p>番外一篇就是這整個分析我丟到 Gemini 與 Claude 去分析，兩個雖然都能說這是 Phishing，但 Claude 就比 Gemini 分析的要好的多，Gemini 保守到假設什麼資料通通都送出去的等級給建議，相對的 Claude 就很好的分析出哪些是風險高、哪些風險低，然後也分析出這個惡意網站是躲在 CloudFlare，叫我去跟 CloudFlare 去舉報。這一點很重要，因為我直接去跟 CloudFlare 舉報後，6 小時內這個網站就被停權了。</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zREfO-YOgMwh6kmuNf6_Lw.png" /></figure><p>至於 Google，對 Google Safe Browsing 舉報了好多次，也請朋友一起舉報，整整 24 小時過去也沒有反應，照樣可以被搜尋到。我問 Gemini 要跟誰舉報，它居然叫我先去跟 TWCERT/CC 去舉報。</p><h3>偷走了哪些東西</h3><p>由於惡意程式碼全部都被多重的 Obfuscate 所以很難分析它到底傳出什麼，我也不是專門的實驗室可以做個假環境去觀察，但判斷底下幾種東西應該是出去了:</p><ol><li>所有的 Web Session Tokens</li><li>.env/.ssh/.aws 的內容</li><li>VPN scripts 包含 VPN endpoints</li><li>keychain-db 包含舊的密碼資料 (如果你沒升到 iCloud), Certificates (尤其是 Apple Developer), SMB 服務的密碼</li></ol><h3>事後的補救</h3><p>事情發生的太快，只能料敵從寬的方式來解決，底下是我的清理清單：</p><ol><li>整個 macOS 重灌</li><li>變更所有 Web Session 還在的雲端服務的密碼，並且撤銷掉所有 macOS 的登入</li><li>撤銷並更換所有的 API keys 尤其是 AI 服務的</li><li>撤銷 .ssh 內所有 keys 在雲端上的權限，並且全部改用 Secretive</li><li>撤銷所有 AWS 產出的 Access Credentials</li><li>撤銷並更換所有的 VPN certificates</li><li>撤銷所有的 Apple Developer 的 certificates</li><li>變更所有 SMB 服務的密碼</li></ol><h3>安全的又是哪些</h3><p>底下的部份經判斷應該是安全的：</p><ol><li>同樣的雲端服務但是是別的裝置的 Session (但有些雲端服務一旦你改密碼所有的 session 都會一律撤銷)</li><li>放在有 Master Passphrase 的密碼管理器的密碼與 TOTP Key，我是使用 Zoho Vault 以及 iCloud 的 Password ，這兩個與 Chrome Safebox 不同，無法被一鍋端，所以會安全很多。</li><li>所有 Web Session 已經過期的雲端服務。</li><li>使用 SSO 登入的服務 (例如 AWS Identity Center)</li></ol><h3>我的建議</h3><p>經過這一次事情，底下是我給各位認真的建議</p><ol><li>安裝任何東西，只要是透過 shell 方式安裝的，絕對要小心小心再小心。能夠用 Homebrew 這類管道安裝的就儘量透過這些安全的管道去安裝，只要看到不熟悉的，先問是不是有 brew 安裝法，shell 安裝法能避免就避免。</li><li>絕對不要用瀏覽器的密碼管理器來幫你管理密碼，尤其是 Chrome，Chrome 的 Safebox 非常容易被人一鍋端走。Safari 用的是 iCloud 所以相對安全。最好是用專用的密碼管理器例如 <a href="https://www.zoho.com/vault/">Zoho Vault</a> 或 <a href="https://1password.com/">1Password</a> 這些有 Master Passphrase 或 TouchID, FaceID 擋在前面的管理器，我自己是用 Zoho Vault ，所以這次沒有造成密碼事件。ps. Zoho Vault 記得要選用 US Data Center。</li><li>雲端服務如果有提供 passkey 就要使用，有 MFA 多重驗證就一定要打開，這種外洩事件一旦發生速度都很快，有這些多重保護，可以免除很多麻煩。</li><li>SSH 金鑰建議用 <a href="https://github.com/maxgoedjen/secretive">Secretive</a> 這種以 Agent 的方式來做，而不是存在本機電腦上的 Private Key</li><li>建議可以安裝 <a href="https://objective-see.org/products/lulu.html">LuLu</a> 這類防火牆，萬一有偷傳發生，第一時間可以幫你擋下。不過 Phishing 一旦請進來你還是得重裝 OS。</li><li>電腦要有好的備份機制，要做到可以當機立斷馬上重灌也不會有資料損失的程度。我的做法是資料全都有 NAS 在後面做同步及備份。</li></ol><h3>充滿險惡的環境</h3><p>各位試想一下，如果有一個惡意的駭客，刻意寫一個小工具，然後透過重重的包裝及各種阻斷掃描的方法，即便工具本身做成無害的東西以偽裝成無辜的第三者，再搭配到處去寫介紹文，再用間接的方法引導你自己走錯路，你就很可能這樣中招了。除了使用者自己的警覺外，最好的保護還是來自於平常的習慣，不要依賴單一的保護，而且要保持警覺心，這是我給大家的忠告。</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=39f79565d10d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>