July 27, 2026
Clean Code, dirty drama
The Author of Clean Code No Longer Reviews AI-Generated Code
Clean Code legend says he won’t read AI code—and the internet is losing it
TLDR: Robert Martin says he now lets AI write code without reviewing it line by line, trusting heavy testing instead. The reaction was a mix of mockery, suspicion, and reluctant agreement, with critics calling it a convenient reinvention and others saying this may simply be how software gets built now.
Robert Martin, better known to many programmers as the author of Clean Code, has tossed a match into an already blazing artificial intelligence debate. In a tweet, he said he now lets AI “agents” write the code and doesn’t read it, relying instead on a fortress of tests, quality checks, and rules to keep things safe. In plain English: he’s saying, “I don’t inspect every brick anymore—I judge the house by whether it survives every storm I can throw at it.”
And wow, the crowd had thoughts. Critics instantly pounced, with one of the sharpest replies arguing this is perfectly on-brand for the man they blame for years of bloated, over-complicated software style. Another commenter called it a career pivot with suspiciously perfect timing, basically accusing Martin of switching from teaching humans how to write code to teaching companies how to supervise robots writing it. Ouch. The jokes came fast too: one user mock-announced a fake sequel called Clean Vibe, which honestly sounds one publishing contract away from being real.
But not everyone was purely mocking. Some saw a serious point: maybe “clean code” in the AI era is less about beautiful writing and more about clear instructions, strong tests, and catching mistakes early. The biggest fear, though, was chillingly simple: if the instructions themselves are wrong, AI can confidently build the wrong thing—and even produce tests that “prove” the mistake is correct. That’s the real drama here: has software craftsmanship evolved, or has everyone just agreed to stop looking under the hood?
Key Points
- •A Hacker News post highlighted a tweet from Robert C. Martin, author of *Clean Code*.
- •Martin said his current strategy is not to read code written by his AI agents.
- •He said he uses strict constraints around the agents instead, including unit tests, Gherkin tests, QA procedures, quality metrics, mutation testing, and test coverage.
- •Martin stated that these checks allow him to take advantage of agent productivity while maintaining confidence in the output.
- •His confidence in the generated code is based on it passing a broad set of tests and validation procedures rather than manual code inspection.