I fixed a bunch of nitpicky PR comments just by saying "PR 69 has some comments, fix the issues" to Claude.
It opened the PR automatically, took the comments and applied the fixes (renaming a few things, simple refactoring and added a few unit tests).
Effort from me: maybe 5 minutes going through the PR comments and deciding they were super simple monkey work and 10 seconds typing instructions for Claude.
Then maybe 5 minutes of checking the work, commit and push.
All of this could've been done with a Github integrated Claude tbh without me in the loop more than accepting the changes one by one.
> maybe 5 minutes going through the PR comments and deciding they were super simple monkey work
There’s an old joke about a mechanic banging a hammer on a car and charging $1000 for it.
Man us engineers severely undervalue each other’s worth huh. One thing all this AI hype has taught me is I’ve started to be very conscious of what my value is. It’s a hard habit to break.
I've noticed how much of my daily work is just busywork that can be automated today.
I kinda enjoy it, I can focus more on planning, thinking, networking and solving actual problems rather than typing 1000 lines of unit tests by hand and then doing a stupid copy-paste mistake somewhere when I try to speed things up and spend 2 hours debugging it. (True story)
It opened the PR automatically, took the comments and applied the fixes (renaming a few things, simple refactoring and added a few unit tests).
Effort from me: maybe 5 minutes going through the PR comments and deciding they were super simple monkey work and 10 seconds typing instructions for Claude.
Then maybe 5 minutes of checking the work, commit and push.
All of this could've been done with a Github integrated Claude tbh without me in the loop more than accepting the changes one by one.