Robert Hughes

Been going back and forth on this. Right now my setup lets Claude Code run migrations and seed scripts directly against a local dev database, but I've kept it firewalled off from ever touching staging or prod, even read-only. Everything that reaches prod goes through a PR I review manually.

Saw a comment in another thread here about a subagent going off-script and targeting its own orchestrator, which made me second-guess even the local access I've been giving it.

Where does everyone else draw this line?
Is read-only prod access ever okay in your setup, or is that a line nobody here crosses?
And for the people who do give agents broader database access, what made you comfortable enough to do that?

i've been shipping for 8 years. every resume i've ever sent has one line at the top that i thought was my strongest credential.

no interviewer has asked me about it once.

Six weeks in, and I still haven't touched a desktop. Everything architecture decisions, debugging, database migrations, the whole thing happens through a phone screen and an AI agent doing the actual typing.

What's strange is how different the failure modes are from what I expected. I thought the hard part would be not knowing how to code. It wasn't. The hard part was learning to think in prompts instead of logic how to describe a bug precisely enough that the agent fixes the real problem instead of patching the symptom, three times, in three different wrong places.

I keep wondering if this is a preview of something bigger, or just a novelty that works because I happen to have a lot of patience and a phone that doesn't die. Curious what others here think: for the people who've built with AI agents has it changed how you think about programming as a skill, or does it just feel like outsourcing typing?

And for anyone who hasn't tried building this way yet what's stopped you? Time, trust, or just never having a reason to?