No self-reporting. Only code. Built a platform that tries to infer what you know from your actual work. Comments URL: https://news.ycombinator.com/item?id=47815869 Points: 1 # Comments: 0
150 applications. One offer. Each application took 5+ manual steps. Separate tools, separate tabs, separate sites — none of them talking to each other. Generic output. Over an hour per application. Paste a job description — or pull it from any job site with the Chrome extension — and five AI agents run an orchestrated pipeline in under 30 seconds: analyzing the role, scoring your fit, researching the company, writing a targeted cover letter, and tailoring your resume to the role. Sequential where it needs to be, parallel where it can be, each agent's output feeding the next. Also includes a dashboard to track every application. And tools for everything around it: interview prep with mock sessions, salary negotiation, job comparison, follow-ups, thank you notes, and references. Runs on your machine. No subscriptions, no data stored on our servers — just your own Gemini API key connecting directly to Google. Comments URL: https://news.ycombinator.com/item?id=47815326 Points: 1 # Comments: 0
Article URL: https://bizlect.com Comments URL: https://news.ycombinator.com/item?id=47813990 Points: 1 # 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://www.podwarden.com/ Comments URL: https://news.ycombinator.com/item?id=47812131 Points: 3 # Comments: 0
Article URL: https://iamalex-afk.github.io/human-os-patch-33-protocols/ Comments URL: https://news.ycombinator.com/item?id=47810895 Points: 3 # Comments: 0
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
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
Article URL: https://reindernijhoff.net/2026/04/eigen-six-months-later/ Comments URL: https://news.ycombinator.com/item?id=47791449 Points: 2 # Comments: 1
Article URL: https://ghostmeet.sshlab.dev/ Comments URL: https://news.ycombinator.com/item?id=47780614 Points: 3 # Comments: 0
Six months ago I got tired of being woken up at 3am for bugs I already knew how to fix. So I built Helix. Bug hits Sentry. A multi-agent pipeline kicks off. QA agent writes the failing test first (TDD). Dev agent writes the minimum fix, runs the full suite, opens a PR. You get a Slack message with one button to approve. Crash to merged PR in under 10 minutes. Comments URL: https://news.ycombinator.com/item?id=47776525 Points: 2 # Comments: 0
Article URL: https://github.com/bolnet/agent-memory Comments URL: https://news.ycombinator.com/item?id=47773981 Points: 2 # Comments: 0
Article URL: https://spectrelang.org Comments URL: https://news.ycombinator.com/item?id=47772806 Points: 3 # Comments: 0
Article URL: https://hergimuci.gumroad.com/l/ai-agent-blueprint Comments URL: https://news.ycombinator.com/item?id=47772707 Points: 1 # Comments: 0
As I got older I found myself buying games that just ended up sitting in my backlog. I just couldn’t get into the mood to play them. I realized there was a disconnect between what excited me and what actually fit my energy levels and time. That’s what led me to build Slated.gg. Most game recommenders look at games from the outside (genre, themes, etc.) What mattered to me was what’s it like to play. How much focus it needs, what type of thinking it requires, can I play in short bursts, is it emotionally taxing, etc. I built a game recommender around experience and landed on 30 experiential dimensions (so far). I created an AI pipeline that researches and analyzes games by those dimensions and stores them in a vector space. Think sentiment analysis, but for player experience. The recommender then finds similarities by geometric (Euclidean) distance based on other games you like. When I mapped 297 games, it was really interesting to see what clusters emerged. AAA games cluster tightly together, exposing how they are built with the same formula. The edges are more interesting and surface games with real differentiation. These tend to be the kinds of games players have a hard time finding good recommendations for. Visualization and methodology: https://slated.gg/map Try it: https://slated.gg/discover-games Happy to answer questions about the approach or where it falls short. Comments URL: https://news.ycombinator.com/item?id=47766624 Points: 1 # Comments: 0
Article URL: https://github.com/shane-farkas/memento-memory Comments URL: https://news.ycombinator.com/item?id=47765140 Points: 3 # Comments: 0
Article URL: https://nous-lang.org Comments URL: https://news.ycombinator.com/item?id=47762250 Points: 1 # Comments: 0
Article URL: https://github.com/22or/2nnel Comments URL: https://news.ycombinator.com/item?id=47761439 Points: 1 # Comments: 0
本帖使用社区开源推广，符合推广要求。我申明并遵循社区要求的以下内容： 我的帖子已经打上 开源推广 标签： 是 我的开源项目完整开源，无未开源部分： 是 我的开源项目已链接认可 LINUX DO 社区： 是 我帖子内的项目介绍，AI生成、润色内容部分已截图发出： 是 以上选择我承诺是永久有效的，接受社区和佬友监督： 是 以下为项目介绍正文内容，AI生成、润色内容已使用截图方式发出 回应一下预告: 成功开通 4 个 Claude Code Max 5x 项目地址： github.com GitHub - lsdefine/GenericAgent: Self-evolving agent: grows skill tree from... main Self-evolving agent: grows skill tree from 3.3K-line seed, achieving full system control with 6x less token consumption - lsdefine/GenericAgent 长程任务拳打OpenClaw， Token消耗脚踢Claude Code hahahaha，目前我们实验组日常就是使用GenericAgent的。贴个最显眼的结果放在开头！！！！ GA VS Openclaw & Claude Code区别: 仅仅 9 个原子工具，对比 CC 53个内置工具，Openclaw 22个内置工具。GA有更准确的工具调用，高效完成原子工具的MVP 四层分级记忆 | 真的是token开销的杀手锏hhh，具体实现可以代码中看。 举个例子: “Hello”，GA 的 prompt 长度是 2298 tokens，Claude Code 是 22821，OpenClaw 是 43321。 反思驱动自进化：任务完成后会自动蒸馏成一个SOP | 这个实现起来不难 结构化浏览器提取 | GA最大的杀招！！！！！ 拳打Playwright, 目前市面上最好的Agent Browser工具。Token开销极大程度降低的同时，精度极高!!! 后续会将该功能上线到skills.sh中 WebCanvas 比 OpenClaw 高 11.2 分，token 用量是它的 1/4。 再贴一点打榜记录吧，其实打榜不重要，更重要的是实际体验： Tips | 这里才是给L站看的内容: 目前GenericAgent可以过 CRS的客户端检测 ，小批量测试过Claude Code的CRS反代 [OK 但还是不建议大家用，Claude Code检测一直在变，我们虽然模拟了客户端，对System Prompt做了处理，解析了Claude Code源码，但不能保证100% 关于测试：我们是测试了某闲鱼上的crs Claude Code Max中转， 50刀成功用完，没有出现报错]。 踩坑：GA不会像openclaw那样傻傻的秒封(虽然前期测试过CRS的时候，报废了好几个Claude Code Max 20x账号，真的就是输入一个hello就被秒封) 目前还在努力适配Anyrouter大善人，能正确返回429，503错误。理论可用，但主要大善人最近一直不稳定，无法做到实际测试缓存优化是否有效。 支持最新的ampere.sh渠道，目前v1/openrouter会有OAI缓存不足的问题，GA中也进行了适配hhh，努力适配L站看到了所有渠道。哭泣，还被Ampere.sh坑了200+，不过fast真的好快。 缓存优化: GA适配了OAI和Claude Code两种接口的缓存优化策略 内置JS逆向skills，Codex破限skills，打野skills，Warp rotate Skills，无限Freemail Skills[绕过api_key], 无限socks skills。 但是上述skills正式仓库中并不会包含…后续可能L站会发焚决.zip。[其实还有注册机和2api skills hhh] 哎，写到凌晨4.17，希望各位大佬明早看到能支持一下吧… 我要摘取你们的小星星和大拇指！！！ 对啦，马上会有一个新版本，加了一个小功能，预告一下哈哈哈，可以猜猜是什么： 5 个帖子 - 5 位参与者 阅读完整话题
I built Darkdrive, a small self-hosted cloud storage app. Files and filenames are encrypted with AES-256-GCM before hitting disk, so the filesystem contains only ciphertext. The encryption key is not stored in one place on the server at rest. It’s split between: - a session share stored server-side - an encrypted cookie stored in the browser Both parts are required to decrypt files. A compromised cookie or session file alone yields nothing. The core encryption layer is published for review. The full application is not open source at this time. https://core.darkdrive.de/public/a7c3222a5c6e12bef0648266/cr... I chose server-side encryption over fully client-side systems to keep it simple to host while still allowing features like previews, thumbnails, and easy access across devices. Darkdrive is a single PHP application that runs on shared hosting with minimal setup. It encrypts filenames and directory names alongside file contents, and uses split-key server-side crypto. Happy to answer questions about the threat model or implementation. Comments URL: https://news.ycombinator.com/item?id=47752643 Points: 1 # Comments: 0