Sabitlenmiş Tweet
Timo Lins
848 posts
Timo Lins
@timolins
Product Design at ▲ Vercel ✧ Co-founded @Splitbee, acquired by @Vercel ✧ Creator https://t.co/Od2BhLZoa7
Vienna, Austria Katılım Aralık 2009
865 Takip Edilen8K Takipçiler
@giuseppegurgone I have a skill that lets codex talk to claude directly. I then say "confirm with claude until both of you agree", which work amazingly well.
English
@thsottiaux I'd love to have a "pause here" feature for queued prompts. Or some other way Codex doesn't automatically go to the next queued prompt.
Some of my queued prompts are about planing, which doesn't work well if it it just goes of to the next queued prompt when done.
English
@sebastienlorber @mrousavy @jaredpalmer They need to flag commits like that that include tons of invisible Unicode characters.
English
@mrousavy now sure how but they also found a way to make it harder to notice in GitHub diff view 🤔 cc @jaredpalmer
English
@mrousavy Not just crypto. It's a full blown backdoor which uses Solana as communication layer.
I deobfuscated the payload here: #48PsoW8c3L2dFhPHecBj9MQRbQs2qa8W3hV33Qx5e6Qc" target="_blank" rel="nofollow noopener">privatebin.net/?8331d963e19d2…
It has a region gate to exclude Russian targets. 🤔
English
after 2 years, this was my last week @vercel. when i first started, @v0 was still just a concept — it's since evolved into something much bigger than that, and i'm so grateful to have had a hand in making that possible 🖤
i'm thankful to have worked with such talented, humble people to ship some truly amazing things. some final details from the last few weeks:
English
Hey @rauchg this email is kinda useless on it's own... since on the surface it just shows "more traffic"... yeah cool
BUT, when you actually click into it, it says "your function time went from 500ms-2s -> 1-2min"
NOW THAT WOULD HAVE BEEN SUPER USEFUL IN THE EMAIL 😅
GIF
English
@peduarte mine runs inside a macOS VM, with it's own GitHub account. So I'm still in control what is has access to. I generally use it a work assistant. i'm not hooking up all my personal life things.
x.com/timolins/statu…
Timo Lins@timolins
I honestly can't believe how smooth this runs. clawdbot running inside a full macOS VM – while using only 3 gigs of RAM. 🤯
English
@timolins full FS access where? on your primary/personal machine? on a vm? on a dedicated mac mini? i could see it being useful if it's on your primary machine, but that comes with a bunch of downsides :D
English
@timolins thanks. this is what i've been after! so:
- cron jobs
- good memory
- learn from your requests (memory related)
do you care about any of the remote capabilities?
English
Timo Lins retweetledi
Rumors are true – we've shipped a sidebar at Vercel.
With the platform expanding, it became tricky to find the tab you were looking for. After many iterations, we think we've landed on something that feels right.
Try it. Give it a few days. We'd love to hear your thoughts 🫶
English
@SherryYanJiang you don't get to buy a new mac mini when using the old laptop
English
is there any big difference in using mac mini vs just an old laptop??
Sherry Jiang@SherryYanJiang
this is all i see on my timeline right now
English
I was curious how hard it would be to roll my own analytics service for my personal site.
Turns out, not that hard (to make a shitty v1).
Anything is possible!
English
@MeltedMindz @CdeBurner The gallery is only for off the shelve VMs. macOS needs to be installed, but UTM makes that super easy.
English
@timolins @CdeBurner where did you get the mac os for UTM I only see other distros int he gallery
English
I honestly can't believe how smooth this runs.
clawdbot running inside a full macOS VM – while using only 3 gigs of RAM. 🤯
Timo Lins@timolins
Before you buy that Mac Mini for clawdbot: Get yourself UTM and setup up a macOS virtual machine for it. It's free - and gives your lobster a safe home. And if you end up buying that Mac Mini anyways, you can just copy the .utm file over and you're done. I was surprised how well this setup works, even when giving it just 4GB of RAM.
English
@vishalsachdev @francedot unfortunately not, Apple enforces a limit of 2 macOS VMs per device :(
discussions.apple.com/thread/2560856…
English
@francedot @timolins will try this soon. have a class of 40 students. Is it worth exploring if i can give them each a clawdbot? on a m2 max, 94 gb , 1 tb hd ?
English
Before you buy that Mac Mini for clawdbot:
Get yourself UTM and setup up a macOS virtual machine for it.
It's free - and gives your lobster a safe home.
And if you end up buying that Mac Mini anyways, you can just copy the .utm file over and you're done.
I was surprised how well this setup works, even when giving it just 4GB of RAM.
English
@_boraturan yeah, no AI inference happening on device, but it's still running a full macOS at 60FPS+. I was surprised about it, especially after trying a bunch of lightweight Linux distros first which felt way laggier.
English
@RodgersMag sure, but it's not just about the files. Having actual apps (like a code editor) is also neat, so you can say things like "open that skill we just installed in Zed" and it works. Or watch it browse Chrome.
I don't get the Sandbox part. Isn't the VM also a sandbox?
English
@timolins You could just drag and drop folders/ files and have it work in a sandbox. If you log all the tools llms run you realise that your are one prompt injection away from being hacked. Anyways i have something working that’s why im not convinced
English
@RodgersMag But then you are missing out on the dedicated workspace feel. It's like having a co-worker with their own computer.
English