TV

[TV @Tv64505]

@1__________l1l_ @opencode I can't do it either.

[黑色马里奥｜买美股上币安 @blmario669]

闲着没事，大概追踪了一下 @jheioff 被盗的资产流向 ，大概是约 171 万美元资产，其中约 1,565,980 USDT + 49.96 ETH + 746,475 HSK。

这个转币的人大概倒了好多手，目前归结到了 4 个私人的 EOA 钱包地址中。

0xB88dA2373084e42358E7bBE9b76BdcF734Aaa013
0xFF9c79B3d6151cEc643099f10c15Cbb61E435Fb8
0x1Ab5222021180A674f5FB9F08526474E80043833
0x8CA2EA87Cd281e4feb357Ff1041b3361Cb28F6cF

等于说这个钱要是不转移到其他交易所，大概率是追不回来的。

另外就是，假如这哥们混币器里混一下，再去追踪都难。

结论就是，这个钱，即便是 @Gate 配合，基本也拿不回来了。

所以现在比较麻烦的点在于，如果 Gate 认定这是用户自己的操作疏忽，问题出在用户侧，而不是交易所系统侧，那这个事情就会非常糟糕。

从目前 Gate 的口径来看，他们更倾向于强调用户需要对自己的账户安全承担主要责任，包括密码、2FA、登录设备、验证码、人脸验证等敏感信息的保管。

也就是说，只要后台日志显��相关操作通过了 MFA、人脸验证，或者有历史验证记录作为支撑，交易所通常就会倾向于认为这是用户侧安全失守，而不是平台系统漏洞。

这种舆论情况下，Gate 或许会给出部分补偿（概率也不大），但想要全额赔付，感觉不太可能。

————————————————————

大概追踪了一下资金流向：

总资金：约 171 万美元资产

约 1,565,980 USDT + 49.96 ETH + 746,475 HSK

原始收款地址

0x34660cD71651977D221Ea6fDA506f5D096064133

————————————————————

约 100 万 USDT 路径

金额：约 999,998 USDT + 40.96 ETH

0x34660cD71651977D221Ea6fDA506f5D096064133
-> 0xd4355D99ef1b6F98d3A9AfE8BcA6D0C0852e45e7
-> 0x789172F3Be2f8D75e5074B6726Ac527Aa3E87a35
-> 0xFF9c79B3d6151cEc643099f10c15Cbb61E435Fb8
-> 0x5b6cf102ba75f51133d409668cf03cb0fefb1596
-> 0xB88dA2373084e42358E7bBE9b76BdcF734Aaa013

————————————���———————

USDT / HSK 换币路径

换币金额：565,982 USDT -> 约 319.116 ETH
换币金额：746,475 HSK -> 约 57,916 USDT

0x34660cD71651977D221Ea6fDA506f5D096064133
-> 0xb1EB41492DFC1BD7024286D160efe69a2a6f774e
-> �� MetaMask/Kyber 换币
-> 0x789172F3Be2f8D75e5074B6726Ac527Aa3E87a35
-> 0xFF9c79B3d6151cEc643099f10c15Cbb61E435Fb8

————————————————————

ETH 后续路径

金额：400 ETH -> 354.98 ETH -> 171.308 ETH

0xFF9c79B3d6151cEc643099f10c15Cbb61E435Fb8
-> 0x098f3eC7b8Ce0e1cd684daFc55D46EC00b49f020
-> 0xeb9B692f110dB8BBd0dE326885d4E450e25d0234
-> 0x1Ab5222021180A674f5FB9F08526474E80043833
-> 0x8CA2EA87Cd281e4feb357Ff1041b3361Cb28F6cF

————————————————————

补充中转

0xe7e1b8160bf277bafc8daf32ef0b0a5c2063940e

这是一笔 100 ETH 中转，后面进了：

0x789172F3Be2f8D75e5074B6726Ac527Aa3E87a35

[TV @Tv64505]

@xianzhe9527 完全同意

[Meliasih @meliasiih]

Sekarang kamu bisa menggunakan Claude Sonnet 4.6 dan GPT-5.4 secara gratis di Zyloo.

Secara singkat,

Claude Sonnet 4.6
• Cocok untuk refactor, code review, dan dokumentasi
• Jago membaca codebase/dokumen panjang
• Output tulisan lebih natural

GPT-5.4
• Unggul untuk reasoning kompleks
• Kuat membangun aplikasi dari nol & desain arsitektur
• Sangat baik untuk AI agent dan multimodal

=> https://t.co/qAfE5SS6ps

• Daftar dengan google
• Buat API key

Note:
-This post is not an endorsement or sponsorship.
-Iam Just sharing in case it's useful to someone else.

[🦊 GitLab @gitlab]

What if the question isn't "how do we introduce AI" but "how do we rebuild work around it?"

GitLab CIO Manu Narayan shares how teams are using agentic AI to focus on what matters most, and why tokenmaxxing is the wrong metric.

Read more: https://t.co/jexPJfit3p"

[🦊 GitLab @gitlab]

Are we generating AI code faster than we can control it? 🚨 Get the answer in our AI Accountability Report.

More in the full report: https://t.co/XVhh7S5iF8

[🦊 GitLab @gitlab]

Back at Nasdaq! 🔔 Ian Steward, GitLab's CRO, rings the opening bell on the Nasdaq floor.

From our historic, livestreamed IPO in 2021 to today, one platform for the entire software development lifecycle has never been more essential.

[GitLawb @gitlawb]

do you want to have unlimited and FREE GLM 5.2 access in 24H?

curious what you gonna build.

please register to https://t.co/VEgNeqDDic and send email to:

[email protected] for whitelist

[🦊 GitLab @gitlab]

We surveyed 1,528 developers and tech buyers across six countries with The Harris Poll and found AI is reshaping software development. But how do organizations make it scale?

The full AI Accountability Report drops tomorrow. You don't want to miss this 👀

[Barret李靖 @Barret_China]

大概一个月多前，电脑中了木马，潜伏了一个月才被发现，也导致了我的 X 账号被盗。

让 Claude 和 Codex 逐一分析系统进程时，找到了一个叫做 com\.apple.accountsd.helper.plist 的进程，乍一看还以为是苹果系统的宿主进程，但 Claude 对这个进程提出了质疑。

沿着这个线索，对 accountsd 进行了详细的分析。这个木马首先给系统添加了一个开机自启项，确保可长期潜伏，然后通过 root 权限写了一个守护脚本，每秒探测当前系统是否有新用户登录，一旦登录就通过 AppleScript 脚本切换成当前用户身份，运行一个叫做 AccountsHelper 的程序。

对 AccountsHelper（SHA256：9168cbc45f）也做二进制分析，它的依赖极少，主要职责是从远端加载指令，然后拉起一个交互��� PTY shell，攻击者就是通过这个 shell 来远程操控电脑的。木马执行的每个环节，清理工作都做的特别好，几乎所有日志都被清理干净了。

为了溯源整个中马过程，我分析了将近两个多月的系统日志，仅找到一条可疑指令：有个 curl 操作，下载并执行了一条混淆命令，这也是唯一的线索，解密指令得到了一个远端 IP地址。

后来也从公开情报里找到这个木马，它是 AMOS Stealer 恶意软件家族的变种，之前主要针对 Windows 平台，今年四月第一次在 macOS 上被发现。

从 .zhistory 日志看，curl 前后都是 Claude Code 相关操作，高度怀疑就是 Agent 程序引入了这个木马。

我的 Claude Code 长期是 bypass 模式，所有的命令执行都是直接放过，怀疑是在做软件安装或更新的时候，AI 从互联网找到了一些不安全的资源，下载了这个木马。

这个 AMOS 木马，主要会去扫各种虚拟货币钱包，尤其是浏览器插件钱包。顺带也把我各种登录态 Cookies 全部拿走了，这才导致 X 账号被恶意添加了一个 Passkey。

一般木马进来之后，除了目的性的攻击（例如虚拟钱包转账、文件加密勒索）外，它还会使劲去找各种敏感信息，尤其是浏览器里的登录态、Keychain 里保存的账号密码、聊天工具的登录状态、开发环境里的 .env 文件和各种 token，等等。

我这个教训，大家记住两点：1）所有能够加 F2A 的账号都要加，登录时的二次验证会增加攻击者的门槛；2）AI 在执行各种程序的时候，一定要注意它在执行什么，尤其是安装和更新软件的时候，要先确认再执行。

跟一些安全研究者也交流了这个木马的细节，他说近期这���事件特别高频，攻击者会伪造各种软件的官网，Google 搜出来默认都排在第一位，诱导用户（尤其是 AI）下载木马，防不胜防！

[🦊 GitLab @gitlab]

4 announcements from GitLab Transcend, and the numbers do the talking:

50x faster agent execution with next-gen SCM. 11x faster responses and 4.5x fewer tokens with GitLab Orbit. Full audit and policy controls with Governance for Agents. One annual commitment that actually flexes with GitLab Flex.

Blogs in the replies 👇

[唐柏桥 ( Baiqiao Tang ) @tangbaiqiao]

够了！我已经听够了这些狗屎言论。为什么要给这个刚杀了四万多人的魔鬼政权提供3000亿美元重建基金？！

万斯说，只要伊朗不再发展核武器，他们就能拿到这笔巨款。那么当年他们为什么要反对奥巴马促成的伊核协议和向伊朗提供基金？当年伊朗不是也承诺永远不发展核武器吗？还有，今后伊朗继续屠杀抗议民众，是不是就没事了？原来这场战争的最终目的是给伊朗魔鬼政权颁发杀人执照！如果美国政府用这种方式跟魔鬼打交道，中共拿到杀人执照指日可待。

当初去炸人家，现在又让自己的小兄弟做冤大头出钱帮对方重建。前提只是十年前他们就答应了的不发展核武器，甚至没包括停止镇压人民和释放政治犯！这算哪门子交易？！但愿我们现在看到的信息都是错的，而万斯的这番言论并不代表川普和美国政府。

[Zed @zeddotdev]

Great software always took shape in conversation, not the commit. With agents, the conversation that generates the code is becoming the true source of our software. And Git can't keep up.

So we built something that can. Meet DeltaDB: https://t.co/x0UHntKy9m

[ServiceNow News @ServiceNowNews]

Together with IBM, we're turning decades of trapped enterprise data into fuel for AI.

Legacy systems weren't built for AI—but they hold the data that makes it work.

Our expanded collaboration with IBM helps enterprises modernize aging applications, govern their data for AI readiness, and automate IT operations—without starting from scratch.

Learn more: https://t.co/RqqSD1ur3A

[Kiro @kirodotdev]

You can now run specs and connect GitLab repos in Kiro Web. 🧵

@darkosubotica

[🦊 GitLab @gitlab]

Most enterprise software contracts lock-in spend before teams know how they will use it.
GitLab Flex is one annual commitment that covers platform seats, GitLab Credits, and new eligible capabilities as they become available.. No contract amendments required.

GitLab Flex is now accepting requests for orders. Learn more here. https://t.co/UtsSWl7crS

[🦊 GitLab @gitlab]

Today is the day. GitLab Transcend is live from London.

Bill Staples, CEO, GitLab, is sharing the latest innovations making agentic AI for software engineering faster, smarter, and more cost effective at scale. Plus, hear from Anthropic, Mercedes Benz, Google Cloud, AWS, Cube, and Compare the Market on where software innovation is headed next.

Register for the livestream. https://t.co/rV8tbcnbie

[🦊 GitLab @gitlab]

Happening tomorrow. GitLab Transcend streams live from London. RSVP 👉 https://t.co/rV8tbcnbie

Don't miss the keynote featuring GitLab CEO Bill Staples and CPMO Manav Khurana, followed by a customer panel with @AnthropicAI, @googlecloud, @AWScloud, @MercedesBenz, Compare the Market, and Cube.

Stanford SWEPR scientists will share their latest findings on software engineering productivity in the AI era, and a live workshop will give you hands-on time with GitLab Duo Agent Platform.

[TV @Tv64505]

@TaoRay 限制回流?逻辑是什么?

[🦊 GitLab @gitlab]

From basic code assistance to AI agents running in production, this is how you scale a DevSecOps platform to a 400% ROI.

European technology partner Cube has placed GitLab Ultimate and the GitLab Duo Agent Platform at the center of their entire software development lifecycle. By using GitLab as their single source of truth, they are running custom AI agents in production and connecting external tools like Anthropic's Claude via Model Context Protocol. Operating within strict ISO 27001 guardrails, they have automated repetitive workflows without sacrificing security.

See how Cube achieved 50% faster project setups and multiple daily production deploys. Read the full customer story here: https://t.co/nGr6Esz6pL