Joined May 2008
775 Photos and videos
Failure mode of remote controlling Claude Code on a Mac Mini: system runs out of memory...
112
So Codex company subscription does not have an equivalent to Premium Seat in Claude Code? That's hard for a company to switch even if Codex actually performs better.... a standard seat's usage limit can be easily burnt through by one prompt on GPT-5.5
1
1
304
@weiwei9 retweeted
> Dario: “fuck you” > pentagon: “no fuck YOU” > boris cherny: “here’s 2 new skills for claude code that will help you manage that pull request 🥰” never change
In the next version of Claude Code.. We're introducing two new Skills: /simplify and /batch. I have been using both daily, and am excited to share them with everyone. Combined, these kills automate much of the work it used to take to (1) shepherd a pull request to production and (2) perform straightforward, parallelizable code migrations.
74
271
7,710
589,235
@weiwei9 retweeted
You can’t imagine how fast Chinese humanoid robots are evolving. In just one year, they have evolved from robots to "humans". 2025&2026 Chinese Spring Festival Gala
638
4,186
17,281
2,861,495
How to use ctrl o to expand an old section in Claude Code TUI? It seems to always expand the latest one even if I scroll back @bcherny
3
1
26
17,790
Very interesting tech from fly.io : fly.io/blog/code-and-let-liv… Some amazing features: 1. It takes about 2 seconds to spin up a full Ubuntu VM with common tools installed. 2. It has a durable storage. 3. Auto sleep 4. You pay for cpu/memory UTILIZATION
1
152
I'm still playing with it but have a feeling this is an awesome agent runtime
73
Claude Code on the web is less useful than local because its webfetch gets blocked way more often What's funny is Claude will still talk in a straight face as if it knows everything despite you can see it got blocked like crazy 🤪
89
@weiwei9 retweeted
Jan 12
it's insane how much ramp built - it's like our entire roadmap and they just went and built it for their team
The craft of engineering is rapidly changing. At @tryramp, we built our own background coding agent to accelerate faster. We call it Inspect. It wrote 30% of merged frontend backend PRs in the past week. It’s powered by @opencode, @modal and @CloudflareDev. It runs fully in the cloud, and starts in seconds, letting every builder work at the speed of thought, no setup required. Today, we’re open sourcing the full blueprint so anybody can build their own Inspect. Just give our spec to your current coding agent, and let it build your new favourite.
22
27
1,441
274,904
This tool is so awesome! Instead of asking my team mates "what did you ask Claude" I can just let them send me the whole session from now on! Thanks @simonw ! tools.simonwillison.net/term…

1
117
Re-reading the tweet that coined the term "vibe coding". Apparently if you still do "code review", it's not the original meaning of vibe coding, it's just AI-assisted coding.
There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like "decrease the padding on the sidebar by half" because I'm too lazy to find it. I "Accept All" always, I don't read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I'd have to really read through it for a while. Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away. It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or webapp, but it's not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.
1
100
I don't know if any company should fully embrace true vibe coding for real product yet. It still sounds a bit crazy to me. How do you ensure you specified enough tests? I often skip reading tests or simple scripts nowadays. I guess we're half way there.
50
TUI is a fad It works ok for engineers who live with terminals. It's not going to work for most people. TUI is only popular now because it's cheaper and faster for LLM to process. Once LLM gets fast enough to process GUI, TUI will get back to its own corner.
76
Is OpenCode a Claude Code clone for frugalists? They were bluntantly violating Claude Code ToS. I don't know why anyone stands for them on this matter.
107
Despite how much I don't like Elon Musk and the current X... I have to admit I get most of my AI news and tidbits from X. Threads has a more balanced vibe, but X is still where tech discussions happen.
55
@weiwei9 retweeted
30 Dec 2025
.@1Password browser extension is injecting Prism.js *globally* on every page, which then applies its syntax highlighting logic on all <code> blocks matching [lang=*] regardless of whether it’s meant to be compatible, thus breaking original highlighting. Terrible negligence and even more so that this made to prod while already flagged during beta. Been a user for a long time but this will def push me to an alternative if not fixed soon.
30 Dec 2025
Replying to @evanyou @saltyAom
what’s unfortunate is this was flagged in 1P forums on the beta build over three weeks ago, still made it to prod, and still hasn’t been fixed. really hoping to read a postmortem on this one @1Password 1password.community/discussi…
117
205
2,666
813,277
26 Nov 2024
I just registered for NeurIPS 2024! neurips.cc

1
2
400
@weiwei9 retweeted
We’re Firsthand, the AI agent platform purpose-built for brands and publishers. Find out how and why we’re building the data and agent infrastructure to power consumer interactions in the AI era: venturebeat.com/ai/exclusive… #ai #GenAI #media #Advertising
3
2
5
826
16 Dec 2022
睡了一觉起来发现昨晚要订的机票价格降了400块。。有点捡到钱的快感
1
6
978