3 ms·

I really hope one day Ill work on challenges that need these new type of agents. Currently, I either need a fast agent that does what I want faster than I can

by jwpapi 11mo ago

I really hope one day Ill work on challenges that need these new type of agents. Currently, I either need a fast agent that does what I want faster than I can type it (CRUD, forms, etc) or I need an agent to discuss a plan, ups and downs. Whenever I try to give it a bigger task it takes a lot of time, and often is not what I’ve expected, which might be totally my fault or context specific, but as soon as I’m able to define the task properly I would prefer a faster model as it will be good enough, but faster. I really don’t have problems anymore that I can’t reasonable solve fast enough with this approach. I’ve run multiple gpt-5 codex concurrent sessions in the cloud, but I didn’t accept one thing they did. Eventually thinking through it, reading hack boom is faster than outsourcing the work for 30 minutes + 30 minutes to digest +30 minutes to change..

spruce_tips 11mo ago
100% agree. composer-1 really has been the sweet spot for me of capability, reliability, and speed. i dont ask it to do too much at once, and this approach + its speed, materially speeds my work up. i generally find i get the most out of models when i feel like im slightly underutilizing their capabilities. the term i use for this is "staying in the pocket"
- jwpapi 11mo ago
  Is it available via api? Cant find it on openrouter...
  - spruce_tips 10mo ago
    it's in cursor only
arresin 11mo ago
That’s the bet cursor took with composer 1. It’s dumb but very fast and that makes it better
the_duke 11mo ago
The key is learning how to provide proper instructions. Treat it as a developer that just joined the project and isn't aware of the conventions. Provide hints for the desired API design, mention relevant code locations that should be read to gain context on the problem, or that do similar things. An AGENTS.md that explains the project and provides some general guidelines also helps a lot. Codex can be incredibly strong when prompted the right way.
- jwpapi 11mo ago
  But when I’m at that point. I think either I myself or a faster agent can do the jobs, ergo no need for a long-running smart agent.. This might be in the nature of problems I’m facing in my coding endeavors. I just don’t have any tasks that I cant solve in less than 45 minutes, or the problem is so vague in my head, that I can't accurately describe it to an AI or human. Then usually I either need to split it in smaller problems or take a walk. Since claude 4 I barely wish, omg I wish this agent would be smarter. I still wish it would be faster. But what you described is of course good practice and necessary for smart execution as well.
- ghosty141 11mo ago
  This is generally the right approach imo (when it comes to codex). In my experience Codex is pretty "bad" at spotting conventions or already existing code. Yesterday I told him a feature to implement (maybe 40 loc?) and he 1. did added unnecessary atomics and 2. he kinda reimplemented a function that already existed that he should've just reused. I told him that and he fixed it but these are the things that kinda hold AI back by a lot. It's MUCH harder to read code than to write it, and if he writes the code I must 100% understand it to have the same confidence in it as if I did it myself. And that to me is mentally almost more taxing than doing it myself. If you just let codex write the code while instructing him exactly what you want in terms of logic and architecture it works really well and saves a on of typing.

by jwpapi 11mo ago

spruce_tips 11mo ago

100% agree. composer-1 really has been the sweet spot for me of capability, reliability, and speed. i dont ask it to do too much at once, and this approach + its speed, materially speeds my work up. i generally find i get the most out of models when i feel like im slightly underutilizing their capabilities. the term i use for this is "staying in the pocket"

jwpapi 11mo ago
Is it available via api? Cant find it on openrouter...
- spruce_tips 10mo ago
  it's in cursor only

arresin 11mo ago

That’s the bet cursor took with composer 1. It’s dumb but very fast and that makes it better

the_duke 11mo ago

The key is learning how to provide proper instructions. Treat it as a developer that just joined the project and isn't aware of the conventions. Provide hints for the desired API design, mention relevant code locations that should be read to gain context on the problem, or that do similar things. An AGENTS.md that explains the project and provides some general guidelines also helps a lot. Codex can be incredibly strong when prompted the right way.

jwpapi 11mo ago
But when I’m at that point. I think either I myself or a faster agent can do the jobs, ergo no need for a long-running smart agent.. This might be in the nature of problems I’m facing in my coding endeavors. I just don’t have any tasks that I cant solve in less than 45 minutes, or the problem is so vague in my head, that I can't accurately describe it to an AI or human. Then usually I either need to split it in smaller problems or take a walk. Since claude 4 I barely wish, omg I wish this agent would be smarter. I still wish it would be faster. But what you described is of course good practice and necessary for smart execution as well.
ghosty141 11mo ago
This is generally the right approach imo (when it comes to codex). In my experience Codex is pretty "bad" at spotting conventions or already existing code. Yesterday I told him a feature to implement (maybe 40 loc?) and he 1. did added unnecessary atomics and 2. he kinda reimplemented a function that already existed that he should've just reused. I told him that and he fixed it but these are the things that kinda hold AI back by a lot. It's MUCH harder to read code than to write it, and if he writes the code I must 100% understand it to have the same confidence in it as if I did it myself. And that to me is mentally almost more taxing than doing it myself. If you just let codex write the code while instructing him exactly what you want in terms of logic and architecture it works really well and saves a on of typing.