Meet Limbani

Seems interesting, can I get free upgrade to try it out

Let's go.... Astra CD on it's way

Has anyone explored this idea: Direct steering like how all current harnesses are doing OR Convert steering instructions as part of reasoning to add that block in history directly as if model is steering itself

What the hell is this @GeminiApp . it did return for astra when i did not mention "no fluff. table" but the result was just too buried deep, once i added this then it just didn't even mention astra and understood something else completly

People of PI, Extension for solving ox-alpha stream end error If you're using ox-aplha from any provider then you must have faced it ending the stream mid session a lot. opencode confirmed that this is a problem from the lab so they rolled out a fix for that in opencode, this pi-extension does the same. it keeps the loop going instead of you having to manually type "continue" github.com/Winds-AI/pi-exten…

I feel like OX-Alpha might be the best model to be used with @mattpocockuk's skills anyone so far?

Yo @thsottiaux , this feature when in codex. more details is really useful but i need different type of explanation for different occasions. it would be helpful if we can figure out something for this problem

Whoaaah At least i didn't know that vercel's agent-browser can be used to let agents control electron apps like slack, figma, notion and a lot more. such a simple idea but so useful, i just found out today when i though of reading the skill file as a inspiration for a tool i'm building.

I’m reaching out on behalf of my friend, whose account @Hacket4 was suspended last month. He has submitted multiple appeals, but every response appears to be automated. He believes the suspension was made in error and, to the best of his knowledge, he did not use bots, automation, or any prohibited tools. He was also an X Premium subscriber before the suspension. We’ve tried the normal appeal process repeatedly and even contacted @Premium support, but keep getting redirected back to the same automated process. @Support @Premium, could someone human please help escalate this?

Yoooo, when did they add this animation in chatgpt chrome extension, looks sooo freakin cool!!!!!!

Hi @petradonka sorry if this is not the right medium but i know you are dedicated to warp enough that you won't mind. is there no support for pi harness in warp? for that custom UI and features that it has for other harnesses?

you get about 3$ of codex inference in openai's go plan at this moment

teapawt.org/petradonka/status/2054… Reminder that this article from @petradonka is one of the greatest piece of writing for thinking about how to prompt agents, even if you know all the concepts in your head but having a written up format in this clean format helps everytime when i am working on prompt optimization for any agent and this is still one of my most used article because i give this to my coding agents(codex) also for understanding the concept to craft initial version of prompt based on problem and it works surprisingly well for agents also

Quick codex tip: if 5.5 in codex feels mialigned to inital task then give it codex session reader skill(i made my own skill as per my requirement, you can make however you like it) and tell it to read your previous goal related messages and realign itself. works supringsly well