It is a separation-of-duties argument for code written by AI agents and also reviewed by the same AI agents.
If you ask Agent1 to write code, the PR diff coming out of it should really be reviewed by Agent2 (or 3 or whatever), the reviewer should be a different system than the code gen agent.
The core claim is that a model is a poor judge of its own work — it shares the same priors and blind spots that produced a bug, so self-review trends toward a shallow review rather than catching blind spots.
The system that writes the code shouldn't also be the thing that signs off that it's safe to ship.
On the flip side, our SaaS runs primarily on GCP so our users are fine. But our billing and subscription system runs on AWS so no one can pay us today.
CodeRabbit has GA'd its support for AI code reviews directly in your terminal. CodeRabbit CLI agent works with your coding agent (Claude Code, Codex, Gemini, etc.) and allows you to automate code reviews and bug fix right as you are building in your CLI.
To get started, run this command in your terminal ask your coding agent to run CodeRabbit reviews:
curl -fsSL https://cli.coderabbit.ai/install.sh | sh
If you ask Agent1 to write code, the PR diff coming out of it should really be reviewed by Agent2 (or 3 or whatever), the reviewer should be a different system than the code gen agent.
The core claim is that a model is a poor judge of its own work — it shares the same priors and blind spots that produced a bug, so self-review trends toward a shallow review rather than catching blind spots.
The system that writes the code shouldn't also be the thing that signs off that it's safe to ship.