Article URL: https://github.com/eph5xx/tweakidea Comments URL: https://news.ycombinator.com/item?id=47815028 Points: 1 # Comments: 0
中转进入决赛圈的是 ： DMIT :LAX.AN5.Pro.STARTER 和 搬瓦工的 DC6 求佬给点意见，主要是用ai的接口和做代理。 还有就是选东京和洛杉矶，佬们怎么看？ 15 个帖子 - 10 位参与者 阅读完整话题
Article URL: https://github.com/smol-machines/smolvm Comments URL: https://news.ycombinator.com/item?id=47808268 Points: 119 # Comments: 48
Once subagents start spawning other subagents, basic questions get hard to answer: what is running right now, what tool did it just call, did the child agent actually do what the parent asked. I wanted a way to verify that each agent is doing the work that fits its role, and to spot when a run goes off track. Lazyagent is a terminal TUI that collects events from Claude Code, Codex, and OpenCode and shows them in one place. It groups sessions from different runtimes by working directory, so Claude and Codex runs on the same repo appear under the same project. Features: - Filter events by type: tool calls, user prompts, session lifecycle, system events, or code changes only. - See which agent or subagent is responsible for each action. The agent tree shows parent-child relationships, so you can trace exactly what a spawned subagent did vs what the parent delegated. - View code diffs at a glance. Editing events render syntax-highlighted diffs inline, with addition/deletion stats. - Search across all events. You know a file was touched but not which agent did it -- type `/` and find it. - Check token usage per session. A single overlay shows cost, model calls, cache hit rate, per-model breakdowns, and which tools ran the most. - Watch a run in real time, or go back through a completed session to audit what happened. Please let me know if there's any feature you want! Comments URL: https://news.ycombinator.com/item?id=47805963 Points: 1 # Comments: 0
查看docker相关信息 docker info 1. 查看docker下容器 docker ps -a 2. 启动容器 docker start XXX 3. 停掉容器 docker stop XXX 4. 重启容器 docker restart xx 5. 删除容器 docker rm xxx #强行删除所有镜像 docker rmi `docker images -q` -f #删除Exited的容器 sudo docker rm $(sudo docker ps -qf status=exited) 6. 运行容器 以启动adminmongo为例 docker run --name adminmongo -d -p 18081:4321 mrvautin/adminmongo # 查看docker日志 docker logs -f -t --tail 1000(行数) 4a7fdcc7ab8e(容器名) 7. 进入到容器 我用过的两种方法 1. docker exec -it XXX /bin/sh 2. docker attach XXX 8. 删除无用docker镜像 docker images | grep none | grep -v grep | awk '{print $3}' | xargs docker rmi -f 9. 更新容器重启celve docker update --restart=no XXXXX(你的containerId) 10. 查看docker信息 docker inspect XXXXX(你的containerId) 11. 删除镜像 docker rmi XXXXX 基本这几个命令，docker就可以上手了 6 个帖子 - 3 位参与者 阅读完整话题
6 个帖子 - 3 位参与者 阅读完整话题
之前一直在福利板块白嫖老铁的百炼coding plan 本身的工作内容基本都是根据业务做开发 不需要多深的技术 感觉29的性价比还是很值的： 4 个帖子 - 4 位参与者 阅读完整话题
Start with a FREE instant website for your AI on the open internet, then work with it to build a business that sells specialized datasets, files, premium reports, blogs, courses and more. Comments URL: https://news.ycombinator.com/item?id=47802211 Points: 2 # Comments: 1
Meta An Update on Meta Quest Pricing Starting April 19, 2026, we’ll be increasing the prices for our Meta Quest 3 and Meta Quest 3S headsets. Est. reading time: 2 minutes 新价格 Meta Quest 3S (128 GB): $349.99 Meta Quest 3S (256 GB): $449.99 Meta Quest 3 (512 GB): $599.99 Meta blames RAM shortage for $100 Quest 3 price hike | The Verge 4 个帖子 - 4 位参与者 阅读完整话题
I started EDDI in 2006 as a rule-based dialog engine. Back then it was pattern matching and state machines. When LLMs showed up, the interesting question wasn't "how do I call GPT" but "how do I keep control over what the AI does in production?" My answer was: agent logic belongs in JSON configs, not code. You describe what an agent should do, which LLM to use, what tools it can call, how it should behave. The engine reads that config and runs it. No dynamic code execution, ever. The LLM cannot run arbitrary code by design. The engine is strict so the AI can be creative. v6 is the version where this actually became practical. You can have groups of agents debating a topic in five different orchestration styles (round table, peer review, devil's advocate...). Each agent can use a different model. A cascading system tries cheap models first and only escalates to expensive ones when confidence is low. It also implements MCP as both server and client, so you can control EDDI from Claude Desktop or Cursor. And Google's A2A protocol for agents discovering each other across platforms. The whole thing runs in Java 25 on Quarkus, ships as a single Docker image, and installs with one command. Open source since 2017, Apache 2.0. Would love to hear thoughts on the architecture and feature set. And if you have ideas for what's missing or what you'd want from a system like this, I'm all ears. Always looking for good input on the roadmap. Comments URL: https://news.ycombinator.com/item?id=47793245 Points: 2 # Comments: 0
Claude Code sessions that read a real CLAUDE.md take 30-60 seconds to initialize. Fine once. Friction when you're switching contexts constantly. llm-primer keeps a pool of warm sessions in the background. When you need one, it's already past the startup protocol. Happy to answer questions. The main motivation was that I'm running LLM wikis (llm-context-base) and switching context between different agent roles constantly. The startup latency was killing the flow. Comments URL: https://news.ycombinator.com/item?id=47778707 Points: 1 # Comments: 1
想提前了解这个步骤， 这个应该是电脑弹出来的对吧？正常来说 点了start之后呢？是让你上传身份照片吧？ 那么手机摄像头验证具体是如何验证呢？复制链接或者什么？ 假设我使用电脑，然后弹出，需要手机摄像头认证，能不能发给身份的人，让他去过验证呢？这里会不会有我和他IP，手机定位不一致的问题？ 如果全部交给身份持有者操作，他需要登录Claude账号吗？ 1 个帖子 - 1 位参与者 阅读完整话题
每次 codex 启动的时候都等很久，然后报错： ⚠ MCP client for codex_apps failed to start: MCP startup failed: handshaking with MCP server failed: Send message error Transport [rmcp::transport::worker::WorkerT ransport<rmcp::transport::streamable_http_client::StreamableHttpClientWorker<codex_rmcp_client::rmcp_client::StreamableHttpResponseClient>>] error: Client error: e rror sending request for url (https://chatgpt.com/backend-api/wham/apps), when send initialize request ⚠ MCP startup incomplete (failed: codex_apps) 不知道有没有佬友知道如何解决 3 个帖子 - 2 位参与者 阅读完整话题
场景：做室内定位，需要使用微信小程序的api 扫描附近的beacon rssi信号 api: wx.startBeaconDiscovery 参数：uuids Array 必填 Beacon 设备广播的 UUID 列表 疑问：我是想要搜索附近的beacon信号 为啥这个参数是必填的，我都还没搜索到，怎么知道有哪些uuid？有点反直觉。 没办法 只能传入全量的uuid 但是看到资料说 又不能一次性传入太多？那到底该如何传？ 看到有的说法是这个uuids类似一个白名单的意思。微信官方文档及问答 看了一圈 没看到有用的回答。 有没有做过同样业务的佬 传授下经验。 2 个帖子 - 2 位参与者 阅读完整话题
We started GizmoSauce ( https://www.gizmosauce.com ) about 1,5 months ago as a side project. We're building embeddable website widgets for businesses and agencies. The main thing we care about is that most widgets make sites slower, uglier, or harder to maintain, so we've been trying to build ones that are: - easy to install - fast / lazy-loaded - zero-layout-shift - easier to style so they feel native to the site - AI functionality helps a lot We’ve done zero ads so far. Mostly just shipping and helping organic clients, pushing changes to production, and trying to make the product better every day, and passing integration validations like for Instagram sync. What’s already become obvious is that building the product is way easier than figuring out how to get customers. One thing we're starting to realize is that we probably shouldn't position GizmoSauce as just "widgets." We wanted to create more commercial integrations with larger clients. I mean, different widgets make sense for different client types: - Reviews / chat / hours feel like a better fit for smaller businesses - The new widget we build now - Store Locator feels more relevant for multi-location businesses, retail, franchises, and the agencies that work with them So the GTM idea I keep coming back to is: - Find a brand with an obvious UX gap - build a working demo showing how one widget could improve it - send it to their digital / marketing team For example, we made this Little Caesars demo: https://demo.gizmosauce.com/demos/little-caesars/ Compared to their current store search: https://order.littlecaesars.com/en-us/order/pickup/stores/se... They don't currently have a map flow there, so we mocked up what that experience could look like with a Store Locator. Main thing I'd love feedback on: If you were us, would it be ok, or too strange to go this way? Should we start by selling to agencies, or go directly after multi-location brands and not care that nobody knows us? And does the "build a custom demo for a real brand" approach sound smart for early traction, or too manual? Comments URL: https://news.ycombinator.com/item?id=47772268 Points: 1 # Comments: 0
Article URL: https://github.com/oguzbilgic/posse Comments URL: https://news.ycombinator.com/item?id=47771623 Points: 1 # Comments: 0
Inspired by karpathy's US job market visualizer[0], I started to dabble with the idea of building live dashboards for certain job markets, starting with quant finance. I extract the career pages of pretty much every major quant firm and classify each posting with a lightweight LLM ETL pipeline. The data is updated daily and the full dataset is available as SQLite for anyone who wants to do their own analysis. [0] https://karpathy.ai/jobs/ Comments URL: https://news.ycombinator.com/item?id=47770792 Points: 1 # Comments: 0
I started working on this project while doing R&D on how to run an LLM on a device efficiently. Built this for me and have been using it since January. It's helping me, and I think it'll be helpful for others, too. I'd appreciate any feedback Public beta: https://testflight.apple.com/join/Aj9xnfQu Github repo: https://github.com/moeen-mahmud/remen Comments URL: https://news.ycombinator.com/item?id=47769431 Points: 1 # Comments: 0
Superhuman Starter lenny-1gjl20uy 1 个帖子 - 1 位参与者 阅读完整话题
首先感谢站内大佬 @aaa6 的开源, 方便的可以去给个start。 服务地址 MoonTVPlus 已经接入 linux.do 注册。 新人第一次公益,小破管道,看流量关闭注册哈。 没想到反响还可以,已经300+人了,我先关闭注册哈,看看效果再开 有佬感觉没有来得及上车,那我在开一会，2026-04-14 下午5点，或者是500人的时候关 关门了哈,不知道扛不扛得住,晚上观察一下 111 个帖子 - 94 位参与者 阅读完整话题