I find it interesting that you are on the enterprise plan and are not default willing to pay more for more intelligence. Most people I know who are on the enterprise plan are wishing there existed a 2x intelligent model with 2x price.
My company is going through the exact opposite, so it kinda depends on the company. We are actively encouraging our devs to NOT use Cursor because of how much more expensive it is compared to other tools we have from our calculations and they even considered dropping Cursor at contract renewal altogether due to their costs being higher than other tools.
most tasks I can do better and faster with composer 2
a fellow engineer reported a bug on a code I had written a few months back.
I used his report as prompt for composer 2, gpt-5.4-high and claude-4.6-opus-max-thinking.
composer found the issue spot on.
gpt found another possible vector a couple of minutes later, but a way less likely one and one that would eventually self heal (thus not actually reproducing what we observed on production).
claude had barely started when the other two had finished
also, i don't have a budget per se.
but it is expected that i over deliver if i'm over spending
Same experience. I told my agents "never use FactoryBot, always use fixtures." They followed it most of the time. Then under pressure (complex model, lots of associations) they'd fall back to FactoryBot anyway. The agent "knew" the rule. It just didn't always follow it.
What actually worked: a Ruby script that scans the output for known bad patterns. If it finds FactoryBot in the generated code, the agent can't proceed. No reasoning, no judgment call, just a regex match and a hard stop. The agent fixes the violation, then tries again. Went from 40 patterns to 138 over 98 model runs as new failure modes showed up.
This matches my experience. The bottleneck isn't what the agent knows, it's what the agent can verify. A knowledge base tells it "don't do X", but the agent still has to remember to check. Giving it a tool that returns ground truth works better. The agent calls the tool, gets a concrete answer, acts on it. No memory required, no drift over time.
I’m working with a friend to build an ui around Pi to make it more user friendly for people who prefer to work with a gui (ala conductor). You can check out the repo: https://github.com/philipp-spiess/modern
For those using something like this daily, what key combinations do you use to record and cancel. I’m using my capslock right now but was curious about others
Someone told me the other day I should use a foot pedal, and then I remembered I already had an Elgato one under my desk connected with my Stream Deck. I got it very cheap used on eBay. So, that's an option too.
Scroll Lock is really good key for that in my opinion. If your keyboard does not have it exposed then you can use some remapping program like https://github.com/jtroo/kanata
I have a Stream Deck and made a dedicated button for this. So I tap the button speak and then tap it again and it pastes into wherever my cursor was at.
And then I set the button right below that as the enter key so it feels mostly handsoff the keyboard.
> Putting aside GitHub’s relationship with ICE, it’s abundantly clear that the talented folks who used to work on the product have moved on to bigger and better things, with the remaining losers eager to inflict some kind of bloated, buggy JavaScript framework on us in the name of progress.
This says more about the author than anything else.
That they don't support a nationalistic paramilitary organization that requires its members to be masked and never known by the public so they cant be held accountable? A nationalized KKK is not something to ever support.
I don't care about ICE one way or another, but calling people "monkeys" and "losers" because they're not building a product to your exact specifications is extremely childish.
It seems to me that the real "losers" are the ones spending so much time bitching and moaning about a software platform they don't like.
They aren't "bitching and moaning" they are moving communities and platforms. GitHub is user hostile run by a company with a pattern for that. Alternatives to GitHub exist and supporting them is not "bitching and moaning", it's building and creating. The fact you can't or won't recognize that is telling.
The oppositie is also true. "Caring" about ICE and wanting to flood the US with illegal immigrants is also a political position, a privileged one at that. It's a luxury belief: https://en.wikipedia.org/wiki/Luxury_belief
Not being an American is a privileged position regarding an issue that negatively affects Americans. It's no different than if I, an American, fail to care about, say, Hungary's slide into autocracy. I am privileged to be able to not care about it.
By that definition, "privileged" looses all meaning, as there is an infinite number of injustices in the universe, and no one is affected by all of them.
For those of us building agentic tools that require similar pricing, how does one implement it? OpenRouter seems good for the MVP, but I'm curious if there are alternatives down the line.
Why not just use only async agents? You can fire off many tasks and check PRs locally when they complete the work. (I also work on devfleet.ai to improve this experience, any feedback is appreciated)
If you organize your folders correctly, you could probably have that dropbox folder synced with one of those services. The maintenance isn't too bad once its up and running, probably more hurdles (proxies, etc) if you are accessing outside the home. If you are the only user, you can use tailscale to access your hosted apps when out of the home.
reply