Would you be interested in sharing your findings? I'm currently experimenting with LLM-generated rust and honestly think it works quite well, however I'm looking for ways to improve the "taste" of the agent.

I pushed a gist https://gist.github.com/Larusso/82c9aa8effb3031d149d3b5a1b96...

Thanks :)

When you add features such as telemetry, curl calls in install scripts, and in general build on public cloud infra, everything assumes you have internet connectivity. That assumption is so crucial that it is embedded in everything and touches most software components. When you already have an established product, I imagine changing it to remove the assumption without a full rewrite can be tricky.

- You get an overview in the Github UI for each step and can expand/collapse each step to inspect its output.

- You can easily utilize Github actions that others have contributed in your pipeline.

- You can modularize workflows and specify dependencies between them and control parallel executions.

I'm sure there are more. But the main advantage is you don't need to implement all these things yourself.

For #1, you can output section markers from any software: https://docs.github.com/en/actions/writing-workflows/choosin... (I've only used this feature with GitLab)

Thanks, I didn't know about this!

That second one sounds more like a security risk to me than a feature.

For young engineers, it is a good thing to spend time implementing what you call "bad ideas". In the worst-case, they learn from their mistake and gain valuable insight into the pitfalls of such ideas. In the best case, you can have a technological breakthrough as someone finds a way to make such an idea work.

Of course, it's best that such learning happens before one has mandate to derail the whole project.

Sounds similar to https://github.com/jhaals/yopass.

You're using 5 twos, not 4.