Models write sloppy code that works but isn't maintainable. Our eval is first to measure: would you actually merge this code?

Many SWE-bench-Passing PRs Would Not Be Merged into Main

the real failure mode of uncontrolled vibe coding: your codebase regressing to your worst engineer.

When the cost of a wrong answer is high, a workflow gives Claude independent attempts at the problem and adversarial agents working to break the result before you see it.

Opus 4.8 is around four times less likely than its predecessor to allow flaws in code it has written to pass unremarked.

CompileIQ is not a magic tool that automatically turns poorly-written code into high-performing code. To get the best value from CompileIQ, you need to start with reasonably high-performing code, which then enables the final compiler-heuristics tweaks to take you to maximum performance.

I thought we had a very clear delineation where vibe coding is the thing where you're not looking at the code at all. You might not even know how to program.

Dex Horthy, coiner of Context Engineering and “the Dumb Zone”, publicly retracted his extremely vibe-coding-pilled call 6 months ago and encouraged people to **please read the code**

A common piece of advice for working with AI coding tools is to simply write more tests because if the tests pass, the code is fine.

Claude Opus 4.7 feels like a real step up in intelligence. Code quality is noticeably improved, it's cutting out the meaningless wrapper functions and fallback scaffolding that used to pile up, and fixes its own code as it goes.

Add contacts, live search, full pipeline dashboard – all unit tests passed.

their productivity is affected by the state of the codebase.

Thinking about how you will observe whether things are working correctly or not ahead of time can also have a big impact on the quality of the code you write.

The DSL has a weaker control over the program’s flow — we can’t have conditions unless we add a special step

The false promise of your source code repository is that everything it contains is “good.” To complete your task, just find something that does something similar, copy, modify, and you’re done. Looking inside the same repository seems like a safety mechanism for quality but, in fact, there is no such guarantee.

What makes it good or bad is the quality of the code being multiplied.

Anyone who’s ever worked with me knows that I place a very high value on what ends up checked-in to a source code repository.