Dear Reviewer: I Value Your Time

That may be the only checkbox we need in a PR template now.

I use agents for coding. A lot. I am not talking about that.

I still review every important change with care. I need to understand the implementation, look for sharp edges, and know what I am asking someone else to look at.

That is not what worries me.

What worries me is the loop where someone asks an agent for code, checks that it seems to work, opens a PR, and asks one or two teammates to review it before they have understood it themselves. Then the comments arrive, the agent reads them, applies the patch, and the author moves on.

At that point the author is quietly saying: my time is worth more than yours.

The obvious next step is worse. If humans are just forwarding comments between agents, why not have another agent review the PR, leave comments, approve, and ship? Add a few subagent review rounds before committing straight to main.

Who are we fooling?

I like to think agent-coding is a tool. An amazing tool. Like autocomplete once was. Like Stack Overflow. Like design patterns still are.

People, always people, misuse tools. They did it by copying and pasting Stack Overflow answers without understanding them. They did it by applying design patterns where they did not fit. Agents are not special in that sense. They just make the blast radius larger.

I am not against agent-coding. I am against outsourcing ownership.

I know that may pick fights. Fine.

Asking another agent to verify that a human read the code would be tricky. Forbidding agents then? Hell no! The only useful control I can think of is embarrassingly human:

• Dear reviewer: I value your time, so I read and understood this change before pushing.

If checking that box feels heavy, that is probably the point.