242. Compact Moments 中
The cover photo shows me making coffee at home over the weekend. I got a very nice-looking coffee puck—very flat and clean, compacted thoroughly. My OCD side really loved it, so I took a quick snap to use as the image for the Weekly.
Record the down-to-earth trending technologies seen every week, and publish them here after screening. If you find it good, you can subscribe to this weekly via RSS to get update notifications.
Trending Tools
🎉 Pake: One-click Web-to-Desktop App Packager
https://github.com/tw93/Pake
Pake, which generates lightweight desktop apps from webpages with one click, recently released version V3.4, named “Panda Magic.” It resolved the issue of incorrect display when application names use Chinese, supported multi-instance running, supported hiding close parameters, supported silent start to the menu bar, enhanced blob scene downloading, and fixed issues with using local icon files.
Google Finance released an AI-enhanced beta version
https://www.google.com/finance/beta
Google Finance has released an AI-driven beta version. You can enter the stocks you’re following and then ask questions based on that content. The idea is great; you can try it out. It even helps you find news items you need to see, helping you make better judgments.
New Concept English (NCE) Online Reading Tool
https://nce.luzhenhua.cn/
Found an online reading tool for New Concept English suitable for beginners: NCE-Flow. Click any sentence to start playing with auto-highlighting following. It also supports speed adjustment, continuous/sentence play switching, and breakpoint resume. The overall interface design is very clean, and it even supports downloading the zip package for local use.
SwanLab: Deep Learning Training Tracking and Visualization Tool
Found an open-source deep learning training tracking and visualization tool on GitHub called SwanLab. It provides a platform to track, record, compare, and collaborate on experiments, mainly for AI researchers. Designed with a friendly Python API and a beautiful UI, it offers features like training visualization, automatic logging, hyperparameter recording, experiment comparison, and multi-person collaboration. The visualization effects are very cool.
Simple API call to send push notifications to your iPhone
https://bark.day.app/
Bark is an open-source custom iPhone push notification tool relying on Apple APNs. It’s timely, stable, and reliable, and won’t consume device power. Based on the system push service and push extension, the app itself doesn’t need to be running. Both source code and backend are open-sourced; you can use this capability to write custom notification tools.
GitHub Copilot CLI Preview is out—A Value Choice
https://github.com/github/copilot-cli
Recently, with Claude Code and Codex used up, I thought about the GitHub Copilot $10 Pro membership GitHub gave to open-source students. Just as the CLI preview version came out, I tried it, and it greatly exceeded expectations. The cost-performance is very high, supporting the use of Claude Sonnet 4.5 and GPT-5 models. The overall experience is much better than the extension version of Copilot. Highly recommended for students considering cost-effectiveness.
Just Looking Around
My current AICoding priority selection
I prioritize Claude Code ($20/month)—fast and decisive, like a young engineer. After it’s used up, I switch to OpenAI Codex ($20/month), which is slightly slower but equally excellent. After that, I use GitHub Copilot CLI (free via GitHub, originally $10), which supports Claude and GPT-5 and is a great backup. Finally, my reserve is Grok Code Fast 1 ($25 top-up), with surprisingly high cost-performance. Paired with claude-code-router configuration, the experience is almost equivalent to a lower-spec Claude Code.
Happy 1024
https://cdn.tw93.fun/
1024. I modified some previously hand-written code into this. I wish that the code you write hereafter is always what you want to write—be a hacker and a painter.
