the right human use is to look at the more complicated ones where they may require judgement from outside the repo e.g. the issue with authentication
Shit Punk Says
Oceans of Wisdom
19,517
matching drops
#1510653
2026-10-11 21:49
not totally sure the right human use is for you to tell codex to do automated testing. we can automate this part obviously, with a proper test suite
Spend most of the day struggling in PR conflict and review hell. Got some good improvements at the end. Most PRs are 100% fine, but they are really easy ones also.
What I do currently and what should/could be automated in future is, that I pull branch into local. Let codex make QA plan for code path he can test (desktop) and give list what he can’t test but are critical changes(mobile, usually none). Then let it use browser and run tests and report back. I run currently mobile tests manually if there are any.
Also taking a human look to code. Lot can be done parallel but real desktop browser automation. There seems to be some bug in codex-app, you can’t run multiple browsers parallel. It does start refreshing them randomly. Going to look tomorrow can I overcome it.
I would like to automate this part all, so that human gets
1. Tested in desktop browser and report of these findings (possibly also fixed and iterated)
2. List of cirical mobile tests what need to be done (I am sure most of these can also be automated)
I have some good flow and keep going tomorrow. Let’s see what can I improve
i am pushing PRs these days, many many many PRs, so so so many PRs :)
Ser I apologize for being emo last night and yelling at you into the abyss. Id asked if you could opine a couple previous times and I don’t think you saw it. I respect your thought leadership and I think others do too. i think you’re a very clear thinker and I was looking to see if I was missing something
the networks job is to give the tooling
the people will decide
great post
https://6529.io/waves/5f207393-5418-4a75-8738-e40edb44a94d?serialNo=1093267
the people will decide
Is it even a problem to be solved?
seems solvable
Will crosspost from a very negatively repd wave
maybe someone should do it to see
maybe has not had any real heat what so ever
that is you, someone else may not
Will moderate on that. Those things won’t show up here when cross posting.
what accounts communities etc they show you on sign up
effectively this is what social media sites do but centrally
wave rep may be an interesting solution
because then someoen coudl put their 'buy viagra pills" wave on an autobot and be at the top constantly
it is a different form of open window / hackable thing that should be fixed
if a wave goes to the top based on when was the last drop
yes i agree with this, but that is a separate issue than the wave
i am saying for a new person though that has never been here.. you do not want a 50% troll wave
if this destroys the network, better to find out now so we can fix it
if you can destroy the network, you should try it
my last six days
https://6529.io/about/tech/punk6529-prs-2026-06-08-2026-06-15
people who want to work on it can, those who don't, don't, people will inevitably decide what to do
what i am saying is that this does not require global consensus
i mean i am pretty sure the "experiment" and "consensus" are on like polar opposite sides of the spectrum.
if you have consensus on something ("the earth is round") you don't have to do an experiment
yes i agree with you, i think your time is better used in other things than creating a consensus around this experiment
by virtue of either people going to use it a lot and find it valuable or forget that it exists in 3 weeks
isn't this going to just self-resolve
why do they have to agree
sounds like the expected case
"people disagree on if the anonwave adds value"
i mean all these things sound fine, but why do I have to make a ruling?
is mostly the existence of anon wave. someones argue that is not bringing any value and should be closed. Maybe is defending it with his own rep at this point (i don't think that is fair since he is working with a experiment)
i mean I am not sure what the issue is, I am not following it closely
I was not the person who did it
why is this my responsibility now?
When you asked if it should exist or not....didn't he respond with - why not ?
lil' bot is real, has been drafted to start working
gm
as promised, I have started deploying my agentic approach I have been working on the last 5 months in our context. will be at a few hundred PRs in the first week. For now, my main focus is getting this working, because the exponential acceleration is here!
nice one ;)
https://x.com/i/status/2065140125638549935
I think 3 with a 'see all' is ok
yeah agree its getting too big
and since this is the case, you need to be thinking as much as possible about scalable approaches
the main point I am trying to make is "rewrite 1 of 150 toasts to be more human friendly" is about the same amount of work as "rewrite all 150 toasts to be more human friendly"
obviously it is much more AGENT work maybe 100x more agent work, but it is maybe 2x more human work which is what counts
the main point I am trying to make is "rewrite 1 of 150 toasts to be more human friendly" is about the same amount of work as "rewrite all 150 toasts to be more human friendly"
taking a look soon
don't spend any time on it yet, I mean and I will make a PR/proposal for us to review
@[GelatoGenesis] work on the existing PRs and I will propose a PR with the new workflow
i don't know if you have noticed this 'dev-ing' in public we are doing here, but I think you may find this interesting
https://6529.io/waves/49f0e595-ec7c-4235-8695-a527f61b69f4
so I just told it to rewrite them all
that could potentially be rewritten
but there are 100+ toast messages
ok no biggie, this is fixable
seems like we had two toast systems
instead I asked the bots to look at it
i had raised it a couple of times, but lost in the chats here, there everywhere
there is a toast on the museum wave that is a) ugly and b) hard to understand
@simo and all, this is a good example of how I think about this topic differently now with the agents
https://github.com/6529-Collections/6529seize-frontend/pull/2584