That's a great point
The setup friction is real. Once you're past that... do you find understanding the codebase itself (where things live, why decisions were made) is also painful, or does that come naturally after a few days?
The setup friction is real. Once you're past that... do you find understanding the codebase itself (where things live, why decisions were made) is also painful, or does that come naturally after a few days?
So I am used to looking at a mysterious code base and gradually figuring out how to safely change it. When I run into something that’s particularly dangerous (e.g. how does the auth system work?) I will document it myself, and I love writing “how do I?” runbook/procedural documentation if it is likely I’ll need it in six months or will need to hand something off to somebody else.
In the AI age there is a lot to say for just loading up a project in an agent-enabled IDE and asking questions like “When I do X, the system does Y, why is that?” and “How would I do Z?” and having extended conversations “well, I like D but I am concerned about E?” Or “What if we did F instead?” Even if you write every line of code yourself you mind find an AI coding buddy is more talkative than your coworkers.