"Use this CLI tool and figure it out. Look up this sentry issue using it"

"Add a service that watches for an error in the log. Look at the home assistant MCP data and find my phone, send a notification to it and make it send when there's an error"

Now after it manually does it and makes the code. "Make this into a skill for anytime I paste a URL with sentry inside the message"

But also important: "make that procedure a prompt file so when I invoke it I just pass X after it and it works fully"

Having too many skills for things that are very specific where you can directly invoke it via slash command, wastes context space with the skill headers I find. So I make those prompts instead

> There was an unexpected consequence: By automating the laborious parts of my work, I had, in turn, filled my day with more laborious parts.

> A menial task that took me 5 minutes once or twice a day turned into a menial task that I was doing 10-20 times a day. As I had automated the obvious parts, it left behind the glue work that never bothered me before. Now it was a substantial part of my day, and the context switching was killing me.

I assume that’s not the whole story, but I would have liked to read about the ways it made things better, too.