For me this is where a config layer shines. Develop a decent framework and then let the agents spin out the configuration.

This allows a trusted and tested abstraction layer that does not shift and makes maintenance easier, while making the code that the agents generate easier to review and it also uses much less tokens.

So as always, just build better abstractions.

I work with a project that is heavily configuration-driven. It seems promising, but in reality:

- Configuration is massively duplicated, across repositories

- No one is willing to rip out redundancy, because comprehensive testing is not practical

- In order to understand the configuration, you have to read lots of code, again across multiple repositories (this in particular is a problem for LLM assistance, at least the way we currently use it)

I love the idea, but in practice it’s currently a nightmare. I think if we took a week we could clean things up a fair bit, but we don’t have a week (at least as far as management is concerned), and again, without full functional testing, it’s difficult to know when you’ve accidentally broken someone else’s subsystem

Now that I've returned to working on the project tonight, I just remembered another failing of our code. (I'm not in any way claiming these are universal problems, just that they are something to be wary of.)

Naming is so incredibly important. The wrong name for a configuration key can have cascading impacts, especially when there is "magic" involved, like stripping out or adding common prefixes to configuration values.

We have a concept called a "domain" which is treated as a magic value everywhere, such as adding a prefix or suffix. But domain isn't well-defined, and in different contexts it is used different ways, and figuring out what the impact is of choosing a domain string is typically a matter of trial and error.

I fully agree. (Seeing how good Figment2 is for layered config in rust is wildly eye opening, has been a revelatory experience.)

Sometimes what we manage with config is itself processing pipelines. A tool like darktable has a series of processing steps that are run. Each of those has config, but the outer layer is itself a config of those inner configs. And the outer layer is a programmable pipeline; it's not that far apart from thinking of each user coming in and building their own http handler pipeline, making their own bespoke computational flow.

I guess my point is that computation itself is configuration. XSLT probably came closest to that sun. But we see similar lessons everywhere we look.

All of that is just code.

Frameworks are just overly brittle and fragile libraries that overly restrict how you can use them.

when do you think we'll get to build real software?

This is a very cool idea. I’ve been dragging CC around very large code bases with a lot of docs and stuff. it does great but can be a swing and a miss.. have been wondering if there is a more efficient / effective way. This got me thinking. Thanks for sharing!

also my experience in using these two models. they are trying to recover from oversteer perhaps.

Don’t stress, its very likely that this problem was vibe coded :) It’s insane how much better Claude Code is compared to alternatives lately.

Great point here, the only thing that feels greedy to me is that these larger companies do not contribute back to the foundational libraries that they are building on, even to a minor extent for ecosystem improvements. Perhaps greedy is a strong word.

i’ve always felt that oss licenses needs to include responsible use terms or something. some orgs dont mind paying for value contributed but you need to provide a structure to do so, even if that is on a voluntary basis.

If anyone from Lovable etc sees these comments, great opportunity for sponsorship where it can make a difference upstream.

Some companies have done this well, at a stage Retool use to sponsor a number of open source libs which greatly helped them with exposure to devs. Surely a better way to spend ad revenue imo.

Please let us know if you launch this open source project, we’d all be excited to use it!

Same here, would love to share experiences. From South Africa.

I’m surprised no one mentions Knex. By far my childhood favorite, and for me, far more creative opportunity when comparing to something like lego.

Yes! K’NEX [1] is based on spatial geometry. You connect plastic rods of different lengths into star-shaped connectors. This creates a "skeleton" or wireframe structure. It is much better than LEGO for building large-scale, open-air structures like roller coasters, Ferris wheels, and bridges. It is superior for kinetic builds. Interestingly, K’NEX has released "K’NEX Bricks" in the past that are compatible with LEGO studs, and some modern K’NEX sets include brick-compatible parts.

[1] https://www.basicfun.com/knex/

Now we finally know where the LLMs learned this style:

> K’NEX isn’t just a building toy—it’s a gateway to limitless creativity!

Haha! Great catch. That specific structure ... "It’s not just a [Product], it’s a [Experience/Metaphor]!" ... is a classic trope of mid-to-late 20th-century American advertising. This style exploded during the "Creative Revolution" of advertising (1960s–1980s). Agencies like Doyle Dane Bernbach (DDB) began moving away from "reason-why" copy (which listed technical specs) toward emotional copy. So it is kind of a time capsule, like a website from the 90's. In professional copywriting, this is called Feature-to-Benefit Transformation. In classical rhetoric, this is a form of Correctio - when a speaker replaces a word or a description with a more powerful one to emphasize a point.

I still think of Capsela a lot. That's my weird pick. Weird spheres that had gears and motors and things. It wasn't fancy, I didn't do that many incredible things with it (vs my LEGOs) but wow, still hits as a very special kind of toy, just because it was the only one that actually did things.

It all depends on how you prompt. and the prompt system you’ve setup.. when done well, you just “steer” the code /system. Quite amazing to see it come together. But there are multiple layers to this.

As per other comments, if it’s making them money, why bother banning it