Here’s “Uncle” Bob Martin’s latest: a reply to someone’s question about how much to trust agentic coding output:<\/p>\n\n\n\n
I\u2019m significantly older than you. I started coding in the late 60s. My current strategy is to not read any of the code written by my agents. That\u2019s the only way I can take advantage of their productivity. What I do instead is to surround the agents with extreme constraints. Unit tests, gherkin tests, QA procedures, quality metrics, mutation testing, test coverage, and a plethora of others. In the end, I have very high confidence in the code they produce because they\u2019ve had to run the gauntlet of all of my constraints and tests.<\/em><\/p>\n\n\n\n Right, first off: “I’m significantly older than you” LOL at that as an argument. I’m Very Old too. Touch\u00e9, Bob! Although to be fair, he\u2019s responding to someone else mentioning their age as a reason for finding adaption difficult.<\/p>\n\n\n\n God, I was so bitchy<\/em> about Bob when I first drafted this post. You have no idea<\/em> how much I\u2019ve toned it down. Reading Bob pisses me off because his bombastic tone masks a lack of any valuable insight. Not reading any <\/em>of the code is the only <\/em>way you can take advantage of agentic productivity? Really? This overly-confident tone lacks any kind of nuance that will enable you to take actionable points away.<\/p>\n\n\n\n And breathe.<\/p>\n\n\n\n Let’s tackle the actual question: your agents have written some code. How much should you, personally, directly interact with said code?<\/p>\n\n\n\n The answer is, as ever, “it depends”. But what does it depend on? There are lots of interesting aspects here, but if you’re looking for one single signal<\/strong>, it’s the degree to which humans will want to directly interact with that code in the future. Yourself included. Assuming that you are human.<\/p>\n\n\n\n