🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Accidentally set NODE_ENV=production and couldn’t install packages. Spent hours debugging with Sonnet and Gemini—no luck. Fixed it instantly with O3. My new debug stack: Cursor + Gemini 2.5 → ChatGPT → Cursor + O3 Sonnet for planning.

One thing I really like about Cursor is that I can use different models to critique each other’s work. I often find small improvements I can tweak after making a big change.

Fixed a memory leak bug with Cursor. Now I’m feeling like a 10x vibes coder.

Codex feels like one of those minimum lovable products — rough around the edges, clearly built fast with minimal design help. No file picker, no autocompletions, no hotkeys, confusing task list, clunky env config… but it just works. Shipped 20+ PRs in 12 hours, zero failures.

i got waitlisted for YC AI startup school. Garry Tan shooting my shot, this is my application: - one of the only 2 women software engineers invited to Cursor top 0.01% user meetup - founding eng at edtech startup working 7 days a week, scaled to 1m+ users - ex-meta engineer and

Bottom quartile investing at YC demo day is better than top quartile venture capital investing

Come to join us! 🙌

How DeepSeek thinks. 中国文化博大精深 Ryo Lu

Wow! Diagrams! Cursor — my wish came true. 🙌

Morph (4,500+ tok/sec) ftw

A day in Shenzhen: In one park, I saw three groups of engineers—one testing robot lawn mowers, another tuning the camera on a new phone, and a third training drones to avoid trees. #buildinpublic

Designing Claude Code would be fun! You can see a UI preview in terminal like this:

I use em dashes the same way as chatGPT.

vibe coding is fun till your app has a 3mins starting time

Grōk

Today, avante.nvim has two updates: 1. Fast apply is supported, using morphllm.com to greatly improve the issue of failure when editing files. 2. Only the moonshot provider is used; you can directly use provider = 'moonshot' for one-click access to kimi/k2 models.

I found that the combination of kimi-k2 and morph fastapply is incredibly effective. Since kimi-k2 is a bit slow, using morph fastapply means that when calling the edit_file tool, kimi-k2 only need to return the path and edit_code without returning the lengthy search_code part,

How to spend 85% less on Claude Code using Morph (4,500+ tok/sec) and Kimi.ai 's Kimi K2 (step-by-step thread 🧵)