/a-unslop-code

Most advice about AI-written code is cosmetic. Drop the emoji from the comments. Rename process_data. Delete the line that says // increment i by 1. All fair, and none of it has ever taken a service down.

a-unslop-code sorts every tell into three buckets and fixes them in that order.

Bugs come first, because they are the ones that ship: a swallowed exception with an empty catch, a method the library never had, a // rest of your code here stub that made it into a commit, an import for a package that is not in the manifest.

Wrong for the job comes second: tutorial-shaped boilerplate in a codebase that is past tutorials, an interface built for one caller, error handling for conditions that cannot occur, code that ignores every convention in the fifty lines around it. This is the bucket that costs you later, and it is the one a linter has no opinion about.

Cosmetic comes last. It is real, it just is not urgent.

The scanner is the second step, not the first. That is the constraint the skill is built around. The loudest tells are structural, and a regex cannot see structure. So the order is: run the code, then scan, then read the diff. A clean scan on code nobody executed tells you almost nothing.

There is an unslop-ignore comment for a tell you chose on purpose, because sometimes the generic name really is the right name.

Ships in the unslop plugin, alongside a-unslop-text and a-unslop-ui. Installing the plugin gets you all three.

/plugin marketplace add imisic/claude-marketplace
/plugin install unslop@imisic

Then the command is /a-unslop-code.

The tell catalog is adapted (MIT) from the vibecoded-design-tells research, a large-scale analysis of what people actually name when they call code AI-generated.