Name Cop: Kill the Bad Names First

Credit before anything else

The method here is Alexandra Watkins', not mine. SMILE and SCRATCH come from her book Hello, My Name Is Awesome. This tool applies her frameworks and claims no authorship of them. If they are useful to you, buy her book. It carries the reasoning, the case studies and the craft that no checklist can hold. The repository is unaffiliated and unendorsed, and it ships a notice file that says exactly where the boundary sits.

I put that first here for the same reason it is first in the README. A tool built on someone else's thinking should say so before it says anything about itself.

How I got here

I have named enough things to know how the conversation actually goes. Someone generates a list. Everyone picks a favourite. Two weeks later the name is in Figma, in a repo, on a slide, and the problem that was obvious on day one is now expensive.

Every naming tool I tried made this worse by handing me more options. Two hundred words is not help. It is the same problem with a bigger surface.

How I thought about it

Kill first, score second. That inversion is the whole design. Generating is cheap and everybody is good at it. The hard part is deleting, and deleting is exactly what a person with an emotional stake in their own idea will not do.

So the order matters: the tool asks what you are naming and who it is for, generates candidates across several different angles rather than one, then runs every single one through seven deal breakers. Anything that trips one is dead on the spot. No appeal, no "but it could work". Only the survivors get scored.

A ranked list is a trap. Ranking implies the bottom of the list is still a choice. It is not. A name with a fatal flaw at position seven is not worse than position one, it is out.

What I actually did

Encoded the frameworks so the agent enforces them instead of describing them. That distinction is most of the value. A model that knows about SCRATCH will happily tell you your name is fine. A model that has to walk the checks in order and report each verdict cannot.

The one place I deliberately broke from the source: my own naming set for these tools obeys a clarity rule that outranks distinctiveness. A stranger has to guess what the thing does from the name alone, with no description beside it. That pushed me away from cleverness and toward name-cop, cart-cop, resume-tailor. Plainer than Watkins would advise, and correct for tools that get discovered in a package listing rather than on a shelf.

Where it's heading

It has done its job when it takes a name away from you and you are annoyed about it. That is the point. Being talked out of a name in an afternoon costs nothing. Being talked out of one after launch costs a rebrand.