Mandy

171 posts

Mandy banner
Mandy

Mandy

@bytecodecrux

bullish on Al tools. bearish on the hype. | engineer × creator @bytecodecrux

शामिल हुए Nisan 2025
26 फ़ॉलोइंग43 फ़ॉलोवर्स
पिन किया गया ट्वीट
@·
your /CLAUDE.md is probably too long. frontier LLMs can follow ~150-200 instructions reliably. claude code's system prompt already uses ~50. that leaves you ~100-150 before it starts ignoring everything. not just the stuff at the bottom. all of it. uniformly.
tweet media
English
1
0
16
8.8K
@·
tweet media
ZXX
0
0
2
29
@·
good thing it's just the web, claude code is fine(as we speak).
English
1
0
2
41
@·
claude hitting me with the rate limit today, anyone else or am i the chosen one.
tweet media
English
3
0
3
217
@·
my favorite underrated command in claude code - /context
tweet media
English
0
0
4
85
@·
@sama @gothburz FDEs can observe but can they refuse?
English
1
0
13
3.6K
@·
@gothburz We will deploy FDEs, and have cleared researchers.
English
32
7
241
95.1K
Sam Altman
Sam Altman@sama·
I'd like to answer questions about our work with the DoW and our thinking over the past few days. Please AMA.
English
7.5K
572
10.4K
7.1M
@·
@paulg confidence sure but also he already made the hard calls before the ama.
English
0
0
2
666
@·
@sama how do you enforce the 'human responsibility for use of force' principle in a classified environment where openai engineers can't audit every deployment in real time?
English
0
0
4
656
@·
> 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
Boris Cherny@bcherny

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.

English
75
275
7.7K
588.9K
@·
@fahd838_ yeah, this is how claude code works. i remember @bcherny saying /claude.md files load lazily, so it won't bloat anything. just keep root tight since that one always loads. x.com/bcherny/status…
Boris Cherny@bcherny

In case it’s not clear in the docs: - Ancestor CLAUDE.md’s are loaded into context automatically on startup - Descendent CLAUDE.md’s are loaded *lazily* only when Claude reads/writes files in a folder the CLAUDE.md is in. Think of it as a special kind of skill. We designed it this way for monorepos and other big repos, tends to work pretty well in practice.

English
0
0
4
40
@·
@bytecodecrux What if claude md just 100 lines explain the project and rules But for each file there is md for context if need it Is this better?
English
1
0
0
64
@·
your /CLAUDE.md is probably too long. frontier LLMs can follow ~150-200 instructions reliably. claude code's system prompt already uses ~50. that leaves you ~100-150 before it starts ignoring everything. not just the stuff at the bottom. all of it. uniformly.
tweet media
English
1
0
16
8.8K
@·
made a video about openclaw and this is what you get for being honest lol
tweet media
English
1
0
6
110
@·
i regret making a youtube video on openclaw. period.
English
0
0
2
125
@·
@iruletheworldmo theo is so right on this. read a study a while back, frontier models start breaking past ~150 instructions and the system prompt already eats into that. cut my /claude.md to ~30 lines and yeah, works way better. x.com/bytecodecrux/s…
@

your /CLAUDE.md is probably too long. frontier LLMs can follow ~150-200 instructions reliably. claude code's system prompt already uses ~50. that leaves you ~100-150 before it starts ignoring everything. not just the stuff at the bottom. all of it. uniformly.

English
0
0
28
8.1K
@·
i don't care when you watch this, now, later, tomorrow. but you have to watch this video, there's way too much going on in ai at the moment and i'd be lost without theo. i'd been doing literally everything wrong, sigh. he runs through some interesting papers on how to use your agents md files, skills, context management. super super useful. watch it now, i've changed my mind, watch it right now. stop watching frame mog videos and watch this at once sir.
Theo - t3.gg@theo

You should delete your CLAUDE․md/AGENTS․md file. I have a study to prove it.

English
69
425
7.8K
1.2M
@·
4o image gen had me hooked when it dropped, thought that was the ceiling. nah i was just sleeping on @grok
English
1
0
2
123
@·
claude my dear claude, sometimes you get on my nerves.
tweet media
English
0
0
2
68
@·
Just flipped the switch. Full access.
tweet media
English
42
1
407
43.7K
Mandy
Mandy@bytecodecrux·
@forloopcodes they laid off the people who understood the code and replaced them with ai that doesn't understand it either. at least it's consistent
English
0
0
40
2.6K
@·
I hate that Microsoft might be vibecoding Windows, but it's inevitable microsoft laid off everyone who knows how c++ works so now they just prompt gpt 5 to fix the codebase. 30% of windows is written by ai. that is why your printer drivers were deleted to make room for 4gb of copilot telemetry they rewrote office in typescript. file explorer and the notification center are now just bloated electron instances that take 3 seconds to render a right click menu the taskbar and start menu were rebuilt from scratch in react just to shove ads and "recommended" bloatware in your face. it uses more ram than world of warcraft did in 2004 copilot is being forced into notepad and paint. they are forcing you to test it in your basic tools windows search isn't looking for your files. it's a bing wrapper designed to sell you a microsoft 365 subscription while you're desperately trying to find a local pdf the widgets section is another bloat that nobody asked for. edge webview was designed to keep your cpu usage high enough that you're forced to switch to linux over all of that, the task manager barely works in the latest updates nobody at microsoft knows what "win32" means anymore. they replaced their support forums with an ai that just tells you to "try restarting" if your kernel panics
tweet media tweet media tweet media tweet media
Elan Ruskin@despair

For that matter, Microsoft Word 2002 used about 25MB of RAM. Now Word uses 10x that much memory to display the same 584kb document. What the heck is it doing to that text now that it wasn't doing before?

English
355
2.7K
20.1K
945.7K

खोजें

@sama @gothburz @paulg @cryptopunk7213 @fahd838_ @bcherny @iruletheworldmo @elonmusk