Let's build a technology so powerful it can end civilization, but don't worry, we'll run it in a sandbox we vibe coded so it can't read your email.
03.02.2026 18:29
👍 184
🔁 22
💬 12
📌 2
Let's build a technology so powerful it can end civilization, but don't worry, we'll run it in a sandbox we vibe coded so it can't read your email.
Deno Deploy gets new features as it gains GA status.
⭐ Runs any node.js app or framework
⭐ Tunnels
⭐ Postgres databases
⭐ Built-in traces, logs, metrics
deno.com/blog/deno-de...
Don't be afraid to open draft PRs that you know will be scrapped. Experiment, iterate.. then open a new clean PR with your learnings when ready.
If you’re wondering why your token costs are so high, or you’re hitting limits much quicker than planned.. check tour mcp config.
I don’t tend to use them much but I had linear and playwright enabled as an experiment and it was 60k tokens before even sending a message.. 🤯
Side projects all the way down
Vitals AI analyzes your facial signals in real time, but never saves video footage. Only encrypted data points needed for measurement are processed, so you get instant, contactless health insights, without compromising privacy! buff.ly/ZudlTzj
#DataSecurity #HIPAA #GDPR
We're trying out react-admin and their "anonymous" telemetry was just used to directly target me on LinkedIn to thank me for our trust in them..
That trust just got obliterated. Admin panels are private secure areas.. why is the good faith tracking being used to upsell their enterprise products?!
Did you look at Ruby Central with RailsConf and think.. hold my beer 😅
All of my homelab machines are powered by 96gb Crucial kits 🙈 let's hope none of them break.. even before the exit these things were up from £200 to £1000?!
AI is having a devastating effect on the world in so many ways.. the chaos has to end soon, right.. right?! 👀
www.pcmag.com/news/rip-cru...
More like N^10 😅 each side quest has its own group of nested side quests
Makes such a difference having a good search engine when you’re using it many many times per day. Google was full of ads and slop, others (I used DuckDuckGo for the longest time) just don’t seem to have good results.. I really hope Kagi stays around
The UniFi score 😅
We tested greptile as a code rabbit replacement and it was shockingly bad, like how are people using this. It's even worse than the built in GitHub copilot reviewer.
Missing basic features, wild hallucinations, intern level suggestions, no project context
We cancelled our trial after just 2 days 😅
These LLM attacks are getting insane. This “AI on by default” is going to cause some serious damage one day
Gemini leaking emails from poisoned calendar invites and now GitHub secrets could be stolen by just opening a pull request 🤯
They actually tried to get me to stay by flaunting call of duty on the cancel screen.. absolute worst game they could have picked to convince me to stay for a price hike 😂
Email confirmation showing cancelled Xbox game pass ultimate subscription
If you’re unhappy about the insane price hike Xbox just unleashed on everyone, then cancel.
Complaining doesn’t affect these companies; they will continue to do it if you continue to pay.
Is anyone hosting a production NextJS app on Cloudflare? We've got many large apps that we're looking for move off Vercel but would ideally like to avoid self hosting or AWS lambda.
I've been using @kagi.com for a few days now and I've realised decades of searching has trained me to completely ignore the first result.. but with Kagi it's usually the result I want most of the time since there's no unrelated ads at the top 😅
So I guess moving our entire frotnend stack and apps off vercel and co just got bumped up the list 🙃 triangle man really outdid himself this time
You’re absolutely right!
That was a quick turnaround, nice! Looking forward to the improvements
Scenario: Last week chalk was published with malware along with many other packages. Let's say people find safe-chain as a solution for this and promptly go to install it.
The lax version would actually pull down the infected chalk version as a sub dependency during the safe chain install 💥
Screenshot of safe-chain package showing 6 dependencies
Currently looking at @aikidosecurity.bsky.social safe chain, a tool for preventing malware in npm packages.. but it has chalk as a dependency with no strict version set in the package.json
Surely a tool like this should have 0 dependencies..
A photo of a coffee cup with the AI fix logo and “would you trust a pigeon?” slogan on the side
Pigeon powered coffee from @theaifix.show today
Still no response from support, and now even our tiny projects are taking 25+ minutes to review small PRs.
I asked about learnings, we've got like 500+ in there now. Could this be slowing things down?
Not sure a zoom call would help, but it would be good to get acknowledgement from support
A screenshot of AI agent Code Rabbit saying "Cancel my subscriptions I'm tired of your issues" in response to a GitHub pull request.
I'm not sure the rabbit approves of contacting support..
Blind trust for pigeons. Always.
I've contacted support with more details. I opened some tiny PRs this morning at UK morning to see if yesterday was an outlier and it was 21 minutes for ~35 lines of code changed 🙈
Thanks for your input, was actually helpful to know you're aware of some of this and working on changes.
That’s interesting. Although we’ve seen slow reviews across all projects, even tiny greenfield experiments.. it’s taking 25 minutes per review today on our large monorepo.. which could take time to download without a cache?
Good to know you’re aware of this and looking for a solution.
Will there be an option to revert to the system you had before GPT-5? Along with the slowness we’ve noticed really low review quality across all projects for a few weeks now.
Currently exploring alternative solutions
