@BronGoaTTT2nd@FcMobileAdda@tutiofifa It's very likely people would do it. There are useless 119s and 118s. For example, Marquez 119 is there and no one would use it.
I don't think this is right.
There would be a lot of overlap between the stars that will bring the total count down. You cannot just add all the stars in your different repositories. Just pointing out the fault in calculation. I don't have anything against the project.
Just found out that the entire @supermemory github has more than 31,850 stars ๐คฏ
if all of them were under one monorepo, we would be in the top 500 repositories of ALL TIME.
The only reason we do them as independent projects is so that we can create useful tools for everyone (Even competitors)
memorybench, llm-bridge, install-mcp, code-chunk, apple-mcp, markdowner, and others COULD have been just part of our codebase... but we open sourced them.
I have no reason to do this but just the love of the game. i love this shit
I have no idea if these are actually serious setups.
For every single feature I develop with AI, I still have to verify every single line of code because you just can't trust the output (yet)
I don't understand how people can just talk with AI, not see the code and build apps.
For anyone curious how I manage hundreds of agents via WhatsApp:
I just reply to them.
My system automatically routes my msg to the specific agent that sent the msg.
They all live on the same machine.
Spawning each other autonomously, on default they report to their parent agent.
I am DMing only with "Commander Claude"
(srsly it's already overwhelming enough..)
However, all agents also have a tool to send me direct msgs.
They nearly never do this, they usually just reporting up the chain of command.
All sub-agent msgs get to a single WhatsApp group.
If I reply to a msg directly it goes automatically to the sending agent.
If I just send a msg to the group, it spawns a new agent with it as a prompt.
(and a system prompt explaining all of this)
I just wanted to run some jobs in parallel with this.
But they all multiply and delegate all the time, I just found myself going walkie-talkie:
*click*
"somebody find me the agent that was working on ___"
* walkie-talkie radio static sound *
And I get a msg back from that agent.
(after the new one I spawned found him, woke him up with its context, prompted him to contact me).
"All agents be advised, the knowledge base was updated with documentation of ___"
And then I see on my own terminal, out of nowhere, in the input of the Claude Code that I am using, somebody is typing:
"BROADCAST FROM COMMANDER PELEG: Check the knowledge base ({path I didn't even mention but the new agent found it and went out of its way to help other agents})"
Voice notes are transcribed automatically so I literally just talk into my phone like a walkie-talkie, walking around, driving, whatever.
The agent hears me, delegate, multiply, however they want, then execute.
The agents build tools and register them in a shared file so the next agent that spawns already knows what's available and can use them.
I sent a voice msg today about making some docs for all these tools they built for themselves.
Then ended up with 86 parallel Opus calls in different tmux sessions generating documents from 3000 conversations transcripts I didnโt even know about lol.
I didn't orchestrate any of this lol.
They all figured it out and delegated on their own.
The whole thing runs all the time, I just talk to them via voice and they do their own thing in the background.
Everything running with dangerously bypass permissions, yolo, no sandboxs.. nothing ever happens lol
(stuff do happen from time to time but itโs all backed up and i like speed so yolo)
This is not any product or an open source project btw, srsly itโs insanely simple you can vibe code your own in 5 mins.
It's just an ugly hacky script with ~400 lines.
Held together by โopus ma man, there is another bug in the serverโ.
But it works and I'm productive as hell with this, who would have thought.
My primary IDE is now WhatsApp lamo
I can't believe how bad chats on @grok mobile app is.
As the chat grows, it cannot even understand latest message. You send something and it replies to your message from 2 days back
hahahaha i fucking love how jensen just glazes the f*ck out of every ai lab
last week: "i love openai, we will invest up to $100B maybe even more"
this week: "NO ONE uses AI better than my boy zuck"
next week: "no one builds SEXIER space data centers than elon"
honestly greatest to ever do it - the goat
In the end, it's about people's preferences.
One wants money and freedom
One wants money and stability
But one could also argue that in this age, even a salaried job isn't stable.
I have a friend making $25K a month cumulating 2 full-remote software engineer jobs :
- working a total of 8h a day
- billing north of $120 per hour
=> total almost $1K a day
and he told me this : "why would I bother being an entrepreneur to deal with all the stress and marketing and uncertainity?"
what would you if you were in his shoes?
Exactly my thoughts. But founders won't care about user experience if they are making more money.
This is probably the way,
> Launch the product without a free trial
> Make enough money to afford providing a free trial