And the “one decision” is “should we do it the right way now or half ass it to fix it later”
2026-06-16 00:32:43
67
✊🏼🏴 0dw01 🚫🍊👑 :
“I mean, I know which one I would choose, but what would you choose?”
2026-06-15 17:33:50
82
cwrp :
There is one caveat I should tell you about that needs your decision. The GSL is there but it's not HJWing the KLO because the EACESS is set to 644. Should I flocate the dorobator with a HAFBK or should we bypass it with a Kressen tuner?
😅 Yeah bro, do what you think is best!
2026-06-16 06:46:18
4
Mountain Banjo :
This is what happens when you start running half a dozen agents simultaneously, you end up being a project manager, and you're just trying to surf the chaos, and not deep diving into any one of the six operations going on.
2026-06-15 19:53:32
23
Aneesh Luthra :
“Just do whatever you think is best”
2026-06-16 17:56:35
4
. :
"create for me a puppet show telling me what is going on"
2026-06-16 03:17:04
8
Auge Eidos :
Remember that these things were trained on our github, stack overflow, and reddit content. That is the statistical indictment of humanity.
2026-06-16 22:01:27
4
saadboy.mp3 :
I made a hook that runs the last message clause sends me through gemma4 locally and it sends back “you know better than I do, keep working” if Gemma thinks the agent can figure it out by itself. NOTHING else worked.
2026-06-16 05:59:03
2
ryan :
“Use your best judgment and make the decision for me”
2026-06-16 04:05:52
2
Eddie HD :
This is where you let Codex review Claude’s work and make suggestions.
2026-06-16 00:06:36
4
mrperfect44550 :
Some AI reasoning/thoughts ends in a Tool call. Tools (cmds) may not have right permission or can lead to security risk by leaking info (or polluting context). Enable thoughts to see what's going on (can also create harness for tool too). Giving the model a larger sandbox should help. Models context always drift if we let it run long enough, it grounds itself by "calling a tool", realignment is necessary (atleast for now) so end results are what user intended.
2026-06-16 00:38:16
1
Nate Ate Late :
/goal after you have Claude chat write the Claude.md file for dangerous continuance auth
2026-06-16 01:58:59
0
5amSelects :
glad i’m not the only one!!! 😅
2026-06-15 19:36:45
1
Breaking Curses :
Yooooo. this is annoyingly accurate
2026-06-16 20:52:16
1
Ozais :
Choose what you recommend for the best quality output 😂
2026-06-16 02:55:53
1
Javi :
I go “give me a toy example”
2026-06-16 02:46:32
1
TURFPTAx :
the reason I spool up interpreter AIs to help manage my dev teams.
2026-06-15 17:22:28
0
k0lt346 :
cc mostly executes, not brainstorm. i get a separate ai to audit it before proceeding.
2026-06-16 06:13:47
0
infinityember.ai :
I just crawled out of bed and it’s 5 am and you ran all night…….
2026-06-16 23:30:17
0
ejay mallard :
“Use Todos. No permission needed.” 😂😂
2026-06-16 19:47:37
0
timestopper20 :
Yep.
2026-06-16 01:52:58
0
The Hovsepian Company :
Might have said this exact line once or twice.
2026-06-16 00:28:19
0
AGTConsultingServices :
🤣🤣🤣 Felt.
2026-06-16 00:10:34
0
Ivory Wilds :
me.
2026-06-15 23:12:56
0
thedogmothermusic :
Omg facts 😂
2026-06-16 13:17:29
0
To see more videos from user @building_out_loud, please go to the Tikwm
homepage.