22, making things that does stuff

Joined September 2025
39 Photos and videos
there's gonna be a day where we look at a model like Fable and think of it like a Haiku and that's kinda exciting.
1
1
30
.@datacurve please release a model pair bench where the plan and build models are different so we can see how well model duos are on coding tasks.
1
20
.@tannerlinsley I'm trying to report a bug regarding Tanstack router but the GitHub issues link for the stackblitz 'file based routes' project is unusable in Firefox and fails to run start command in Chrome.
1
1
66
I am unable to define a component and route in the same file in my Vite Project as shown in the docs without facing:
1
1
41
Great read about the current state of engineering with AI.
1
1
85
notprnvbot retweeted
bawse...
🚨 BREAKING: Active supply chain attack across npm, PyPI, and Crates.​io. Socket detected TrapDoor, a crypto stealer campaign hitting 34 malicious packages and 384 versions and artifacts, with attackers repeatedly pushing new releases across ecosystems. TrapDoor targets #crypto, #DeFi, AI, and security developers, stealing wallets, SSH keys, cloud credentials, GitHub tokens, browser data, env vars, and API keys. Socket detected releases with a median detection time of 5 minutes, 27 seconds. The fastest detection occurred 58 seconds after publication.
1
2
140
ok @googledevs cooked with 'Modern Web Guidance'. better awareness of modern web standards for agents. will be testing it out!
2
92
tell your agent to generate visual plans and reports using html in a untracked .visuals folder, game changing move fr.
77
'big pickle' rn:
1
1
125
*actively death spiraling in the reasoning*
1
110
.@thdxr everything okay with 'big pickle' at opencode? it's been cutting off randomly way too often and speaking a little Chinese
1
138
.@theo explaining harnesses summed up in one picture
Agent harnesses aren't the black magic many of y'all seem to think they are. To prove it, I built one.
1
174
notprnvbot retweeted
Mar 30
Endgame
413
1,544
23,791
2,658,232
having an llm one shot a prototype and trying to match that by hand seems like the best way to start when you're not familiar with a topic. Especially in niche topics, you discover more than it's training data when *you* take a second pass at it w/ docs nearby ofc.
1
76
Once you use 'grill-me' to understand the plan, use 'openducky' to understand the code implemented.
My 'grill-me' skill went viral. mattpocock/skills is up to 9K stars. Quote tweets of it are doing numbers. It's the most useful skill I've written, and I use it even outside of coding:
1
94
highly recommend openducky over reading plans generated by your models for bigger features
1
414