Hello, this is my work where you can move lights in 3D inside a photograph. The project page and has some cool interactive demos to play with. Do check them out and let me know what you think! Comments URL: https://news.ycombinator.com/item?id=47815784 Points: 1 # Comments: 0
Hey y'all, Here's a walkthrough of how I use the 9front distribution of Plan 9 on Windows: https://www.youtube.com/watch?v=IzEa2L_Pgw0 Topics covered: - 9front distribution of Plan 9 running in qemu on Ubuntu inside WSL on Windows 11 - Drawterm running natively on Windows 11 - Running arbitrary commands on Plan 9 from Ubuntu - Mounting a directory from Plan 9 on Ubuntu - Using OpenAI codex LLM with Plan 9 - Editing files on Plan 9 with vscode Comments and suggestions are welcome. Comments URL: https://news.ycombinator.com/item?id=47813252 Points: 2 # Comments: 0
I found myself reaching for SF Symbols' 'Copy Image As…' quite often during agentic design sessions, so I made a command-line tool that the agent can use by itself. It exports Apple SF Symbols as SVG, PDF, or PNG. The vector paths come directly from macOS's symbol renderer. Internally it reaches a private ivar on NSSymbolImageRep to get the CUINamedVectorGlyph, draws into a CGPDFContext, then walks the PDF content stream back out as SVG `d` commands. The output matches what the system draws, rather than an approximation traced from rasters. A few things about it: - Every subcommand accepts `--json`, and `sfsym schema` returns a machine-readable description of the whole CLI. - Symbol enumeration reads the OS's Assets.car BOM tree, so the list of 8,300+ names stays current with macOS updates without a version table in the binary. - Each SVG ` ` carries a `data-layer` attribute, so you can retheme in CSS without touching geometry. It's been saving me a bunch of clicking. Please let me know if you have any other ideas for it. Comments URL: https://news.ycombinator.com/item?id=47812964 Points: 15 # Comments: 3
Article URL: https://github.com/devmegablaster/Notch-Pilot Comments URL: https://news.ycombinator.com/item?id=47807159 Points: 1 # Comments: 0
My laptop sits to my left, where the camera handles Windows Hello. My main display is a separate monitor in front of me. So every sign-in goes: look left at the laptop to get recognized, look right at the monitor to find the "OK" button, click, get on with my day. Apparently this is the intended UX. I wanted signing in to be a breeze instead, something that drifts past without you noticing. So I wrote a small Rust Windows service, Breeze, to do exactly that. When a Windows Hello credential dialog appears (Okta Verify, sudo-style prompts, and friends), it checks whether face recognition is the active method and, if so, clicks OK for you. PIN and fingerprint are left alone. The camera sees you, the dialog quietly closes, you keep working. Matching uses AutomationId and ClassName rather than button text, so it works regardless of the Windows display language. A few notes: - ~2 MB single binary, no runtime deps - The service runs in Session 0 and spawns a helper into the user session via CreateProcessAsUser with an elevated token from WTSQueryUserToken - FocusChanged event -> locate "Credential Dialog Xaml Host" -> if a PasswordField is present, it's PIN mode, skip -> otherwise click OkButton - Does not work for UAC / Secure Desktop prompts. Windows blocks UI Automation there by design. Install: cargo install breeze-wh && breeze-wh install. The second command auto-elevates, so no admin shell is needed. https://github.com/evan-choi/breeze-wh Feedback is very welcome, especially from anyone with a fingerprint reader. I only have a face camera, so that path is still untested. Comments URL: https://news.ycombinator.com/item?id=47805987 Points: 2 # Comments: 0
https://www.axios.com/2026/04/17/anthropic-trump-administration-mythos [!quote]+ Anthropic 首席执行官达里奥-阿莫代（Dario Amodei）定于周五走进 West Wing 与白宫办公厅主任苏西・怀尔斯（Susie Wiles）会面 这是他为解决公司与五角大楼之间激烈的人工智能之争所做努力的一个突破 特朗普政府认识到了 Anthropic 的新 Claude 模型 Mythos 的威力 以及其高度复杂、潜在危险的突破网络安全防御的能力。 为什么这很重要？ 特朗普政府认识到了 Anthropic 新模型 Claude、Mythos 的强大力量，以及它高度复杂——可能非常危险——的能力，能够突破网络安全防御。 1 个帖子 - 1 位参与者 阅读完整话题
Would love some feedback! Comments URL: https://news.ycombinator.com/item?id=47804947 Points: 1 # Comments: 0
2019年初 GPT2还被媒体称为文本生成器（text generator），Anthropic现CEO Dario Amodei还在Openai的时候； 7年后… 2026年初 Claude Mythos 还被称为AI LLM的时候。 7年后… 2033年… 2019 Better language models and their implications | OpenAI Reddit - New text generator built by OpenAI considered too dangerous to release OpenAI built a text generator so good, it’s considered too dangerous to release 2026 X.com - We do not plan to make Mythos Preview generally available… 4 个帖子 - 4 位参与者 阅读完整话题
感觉这个opus4.7就是参照现有模型的分数制作的，每项分数都超过现有最高模型一点（或接近），但又不会高的离谱，这是要学硬件厂家挤牙膏吗 其它厂家发模型也有点难了，数据是要超opus4.7，还是超mythos 4 个帖子 - 4 位参与者 阅读完整话题
opus4.7可能真的是Claude Mythos (Preview)特调版…… 我让它帮我做头脑风暴，之前胡编乱造，然后我让它看原文，它来了这个。 chatgpt不说人话；gemini金鱼记忆。 opus4.7可好，又不说人话又金鱼记忆。 拉完了。 已连夜取消续订…… 2 个帖子 - 2 位参与者 阅读完整话题
I built MCP servers for my oscilloscope and SPICE simulator so Claude Code can close the loop between simulation and real hardware. Comments URL: https://news.ycombinator.com/item?id=47801255 Points: 72 # Comments: 12
the Guardian – 16 Apr 26 My friend keeps sending me unsettling social media videos. How do I tell her... People down the rabbit hole don’t always realise their experience isn’t universal, advice columnist Eleanor Gordon-Smith writes. You might have more luck trying a new tactic [!quote]+ 我 30 年的朋友不断向我发送社交媒体帖子和视频，我要么觉得这些帖子和视频不好笑，要么觉得令人不安。我们住得很远，很少见面，所以我们通过消息应用程序进行交流。我多次告诉她，我更喜欢积极向上或可爱的东西，而且我不关注美国政治。 她的生活很艰难，我理解她为什么花这么多时间在社交媒体上。上周，她每天给我发送多个我根本不感兴趣的视频，其中包括一个女性互相扇耳光的视频。她经常相信阴谋论，直到我反驳它们为止。所有这一切都让我心烦意乱。就好像她不知道我是谁。我没有回复任何这些消息，但她不断发送它们。 陷入兔子洞的人们并不总是意识到他们的经历并不普遍。社交媒体的目的是让消费者认为其他人都分享特定的视觉语言，对他们所看到的内容的态度，以及对如何利用时间的偏好。由于你的其他策略并没有达到你想要的效果，你可能会更幸运地表明，对于她试图向你展示的这个世界，你是一个困惑的局外人。如果此内容是废话，则可以视为废话。 8 个帖子 - 6 位参与者 阅读完整话题
Early in my career I worked on DataLab, the sister site to USASpending.gov before it got merged into it. I then worked on USASpending a bit. Datalab had more of a "for the people" storytelling vibe to it which I liked a lot more and I feel that spirit got lost when we "merged" into USASpending. I built The Public Tab in response to this feeling. My core idea: The district you are in is a more relatable measure of analysis for federal spending for the average person. You want to know what is going on around you. You can more easily "follow the money" this way. I have a pipeline that runs nightly/weekly with data digest from USASpending, SAM.gov, and some other big players in the space. You can subscribe to changes and get updates daily/weekly on spending in your district. You can map that to lobbying, new contracts that pop up, as well as what your rep is voting on. The client libraries are written in ruby and are open source here: https://github.com/govapi-rb You can also check out the API Docs here: https://thepublictab.com/docs/api It is deemed "beta" for now.. I'd love any feedback and to hear what is cool and what is not. Thanks! Comments URL: https://news.ycombinator.com/item?id=47798958 Points: 1 # Comments: 4
Article URL: https://hemmabo-mcp-server.vercel.app/.well-known/mcp.json Comments URL: https://news.ycombinator.com/item?id=47798900 Points: 1 # Comments: 2
Dear diary, this is my story: I'd been sharing MCP configs with other devs at work a lot - templates in shared repos, credentials in Bitwarden, everyone cowboying their own env vars. That's a lot of manual wiring and lack of any real control, so there was already a problem statement forming in my mind. Then three weeks ago I was putting my kids to sleep and reading about Jensen Huang saying every company will run 100 agents per employee, and the math started mathing. That evening I kept thinking about what agents actually need to operate in the real world and eventually landed on the same answer as every spy movie ever: basically, a passport suitable for the mission and clever drop-off locations. So I built STACK. True story. - The passport: a signed JWT (EdDSA) that proves which agent is acting, who authorized it, and what it's allowed to do. Works offline - any service can verify it without calling STACK. Agents can delegate to sub-agents but the scope ever only narrows. Max 4 hops. - The drop-off: is an encrypted handoff between agents. Agent A drops off a package with a JSON schema contract, encryption at rest, and a TTL. Agent B collects it, the custody transfers, and the payload gets deleted. Neither agent needs to trust the other. Just like in the movies! All credentials are KMS-encrypted. In proxy mode they are injected at the network boundary so the agents can make API calls through STACK without ever seeing the raw key. To try it, sign up at https://getstack.run , grab your API key, and connect: claude mcp add stack --transport http https://mcp.getstack.run/mcp --header "Authorization: Bearer YOUR_API_KEY" I want to provide a generous free tier and I hope people get value out of it. Keycard ($38M, a16z) does scoped agent credentials, Descope ($88M) does auth flows, Composio ($29M) does tool integrations. I'm a solo founder in Stockholm without funding, but I'm betting the full control plane is where the market is heading. I may be naive about that, but that's the bet. I like betting. Docs at https://getstack.run/docs . Comments URL: https://news.ycombinator.com/item?id=47795391 Points: 2 # Comments: 0
顶尖模型性能对比以及opus4.7以及Mythos Preview数据表 现在只是用opus4.6都已是惊为天人,问了很多程序员,现在写项目代码基本上纯手写的代码量不到5%,并且大多数都是开始0代码开发了.以后随着模型能力不断提高,看趋势纯软应该很难再有这么多岗位. 苦学几年,难不成还没毕业计算机就要成远古遗骸了吗?之前学过的东西都被ai按在地上摩擦,好不甘心. 大家有什么看法吗? 11 个帖子 - 6 位参与者 阅读完整话题
The open source, self-hosted alternative to Restream, StreamYard, and vMix. Self-hosted multistream studio. Ingest from any source -- OBS, hardware encoders, mobile, browser -- over RTMP, SRT, or WebRTC (WHIP coming soon). Switch between inputs live in a browser-based production switcher. Fan-out to every platform at once: YouTube, Twitch, Kick, Facebook, or any custom RTMP/RTMPS endpoint. One stream in, or many. Every destination, simultaneously. No cloud middleman, no per-channel fees, no limits. Comments URL: https://news.ycombinator.com/item?id=47793158 Points: 1 # Comments: 0
Article URL: https://github.com/impactjo/home-memory Comments URL: https://news.ycombinator.com/item?id=47793061 Points: 3 # Comments: 2
据彭博社报道，Anthropic PBC 计划于未来一周向英国金融机构发布其备受关注的 Mythos 人工智能模型。该公司正在逐步扩大「Project Glasswing」计划的覆盖范围，向特定机构提供该 AI 的早期访问权限，此前该公司发现 Mythos 在识别网络安全漏洞方面具有强大能力。 https://www.bloomberg.com/news/articles/2026-04-16/anthropic-plans-to-bring-mythos-to-uk-banks-within-the-next-week?taid=69e0adaabd318b0001e84068&utm_campaign=trueanthem&utm_content=business&utm_medium=social&utm_source=twitter&embedded-checkout=true 3 个帖子 - 3 位参与者 阅读完整话题
I ran an experiement the past ~6 months. I challenged myself to build and ship 25 projects in 25 weeks. I just launched my 25th project, which is a simple site to encourage others to do their own 25-week challenges. I aimed for the Moon and my goals were: - Grow my newsletter to 10k subscribers - Get 10 million unique page views across all 25 projects - Make at least $2k a month after the challenge What were the results? - ~550,000 unique page views - ~1,000 subscribers to the newsletter - I've only made a whopping ~$1,1000 total (ooof) - A lot of press for my big project channelsurfer.tv - I guess I'm kind of slightly more known as a guy who creates weird stuff Overall, it was a lot of hard work but also a lot of fun. I'm a senior engineer already, but I learned more this past 6 months than probably the last 3 years on the job. Comments URL: https://news.ycombinator.com/item?id=47792341 Points: 2 # Comments: 0