The best experiences I have are those where I can describe what I want done with details. Rather than asking it add toml parsing, I would tell it to exactly which library to use ahead of time and reduce the number of decisions available to the model to make. Some of the most effective use-cases are when you have a reference to give it, e.g. "add x feature the same way as in this other project that is also in the workspace", or "make the changes I made to the contents of directory X in git commit <sha here>, but applied to directory Y instead". In both cases it's a lot of copy/paste then tweaking an obvious value (like replacing "dev" with "QA" everywhere).

I try to give the model as little freedom as possible. That usually means it's not being used for novel work.

> The best experiences I have are those where I can describe what I want done with details.

But that's the hard part! You can only eke out moderate productivity gains by automating the tedium of actually writing out the code, because it's a small fraction of software engineering.

I basically never have agents do work without using plan mode first (I use Cursor). So I'll start with a high-level like: "I want to plan out the architecture for feature X, I think there's a linear project already that we should update once our plan is done, but if not we'll need to create one. Here's a picture of my whiteboard, I think I also wrote some notes in Obsidian a few weeks ago that you should look up. Two things off the top of my head to think about are: should we do X or does it make more sense to do Y? And also, I'm worried about how this might interfere with the blahblah system, so can you do some research around that? <ramble ramble>"

Then it crawls around for awhile, does some web searches, fetches docs from here and there, whatever. Sometimes it'll ask me some questions. And then it'll finally spit out a plan. I'll read through it and just give it a massive dump of issues, big and small, more questions I have, whatever. (I'll also often be spinning off new planning sessions for pre-work or ancillary tasks that I thought of while reviewing that plan). No structure or anything, just brain dump. Maybe two rounds of that, but usually just one. And then I'll either have it start building, or I'll have it stash in the linear agent so I can kick it off later.