--- name: name-cop description: A naming process, not a name generator. Applies the SMILE and SCRATCH frameworks created by Alexandra Watkins in her book Hello, My Name Is Awesome. Builds a brand brief, generates candidates across several angles, kills any that trips a SCRATCH deal breaker, and scores the survivors on SMILE. Use when naming a product, feature, company, side project, or repo, when picking between name candidates, or when a name feels wrong and you cannot say why. --- # Name Cop Names things properly. > **The method in this skill is not mine. SMILE and SCRATCH were created by Alexandra Watkins and > published in her book *Hello, My Name Is Awesome* (Berrett-Koehler Publishers). This skill applies > her frameworks and claims no authorship of them. If they are useful to you, buy her book, which > carries the reasoning, the case studies, and the craft that no checklist can.** Full attribution and the copyright boundary are in `NOTICE.md`. Whenever you explain a SMILE mark or a SCRATCH kill to the user, credit her by name the first time it comes up in a session. The failure mode in naming is falling for your own first idea. A checklist that only ever adds encouragement will not stop that, so this one is built around a kill rule. ## The profile: read this first, every invocation `~/.claude/name-cop/profile.md` holds the current brand brief, the names already killed, and a log of past naming jobs. **If that file does not exist, run onboarding (next section) before generating a single candidate.** A name proposed without a brief is a word, not a candidate. The path is global rather than per-project so the brief cannot end up committed to the repo of a product that has not been announced yet, and so the killed list survives across sessions. A name you rejected on Tuesday should not come back on Friday wearing a new font. ## Onboarding: the brief Run this when no profile exists, when the user runs `/name-cop-init`, or when they start naming a *different* thing. Ask one topic per message, six exchanges. Write the file at the end and show it. 1. **What is being named, and what it does in one plain sentence.** No adjectives yet. If they cannot say what it does in one sentence, that is the real problem and naming will not fix it. 2. **Audience, and what they call the problem today.** Their words for the pain are the raw material for a Suggestive name. 3. **Personality.** Three adjectives for the register the name should carry, and one thing it must not sound like. The negative constraint does more work than the three positives. 4. **Names already in the space, and the names it will sit beside.** Two lists, asked together. First: competitors, adjacent tools, anything close enough to be confused with. That powers the Copycat kill. Second: the **siblings**, the names this one will be shelved next to, in the same product line, repo owner, or profile. Siblings set a house style, and house style can outrank a SMILE score: a name that wins alone can still be the wrong name for a set. Ask for the siblings even when the answer is none. 5. **Availability requirements.** Exact-match domain and which extension, package registry, social handle, trademark class. Some of these do not matter for a side project. Ask which actually do. 6. **Hard constraints.** Length, syllable count, language, character set, whether it has to be pronounceable by a reader who has never seen it. Write the profile in this shape: ```markdown # Name Cop brief naming: one-liner: audience: they-call-it: personality: must-not-sound-like: taken-or-too-close: siblings: availability-needed: constraints: ## Shortlist so far | Name | SMILE | Weakest link | Status | |---|---|---|---| ## Killed | Name | SCRATCH letter | Why | |---|---|---| ## Past jobs | Thing | Chosen name | When | |---|---|---| ``` ## SMILE: the five marks of a strong name Score each survivor 0, 1, or 2 per mark. Ten is the ceiling. State the score, do not hide it. - **S, Suggestive.** Evokes something about the experience of the thing. Not a description of the category. Suggestive means it points, not that it explains. - **M, Memorable.** Sticks by hooking onto something the audience already carries around: an object, a phrase, a familiar association. - **I, Imaginative.** Shows a mind was involved. If it reads as the output of a template, it fails here even when nothing else is wrong. - **L, Legs.** Extends. A name with legs gives you a theme for features, releases, error messages, and merchandise without further invention. - **E, Emotional.** Makes someone feel something in the half second before they understand it. ## SCRATCH: the seven deal breakers Check these **first**, because each one is cheap to check and fatal when true. - **S, Spelling-challenged.** Cannot be typed correctly after hearing it once. Dropped vowels, creative consonants, numbers standing in for words. - **C, Copycat.** Sounds like a competitor, or like the same three syllables every other tool in the category already uses. - **R, Restrictive.** Boxes the thing into its first version. A name that names today's single feature becomes a lie after the second release. - **A, Annoying.** Contrived, smug, or a joke the audience did not agree to. If saying it out loud in a meeting is embarrassing, it is annoying. - **T, Tame.** Flat, safe, forgettable. The category word with an industry suffix bolted on. This is the most common failure by a wide margin. - **C, Curse of knowledge.** Only lands for people who already know the inside reference. Clever to the four people who built it, opaque to everyone else. - **H, Hard to pronounce.** A reader stalls, guesses, or avoids saying it. A name people will not say out loud does not spread. ## Process. Follow this order strictly. 1. **Brief before names.** If asked to skip straight to suggestions, say no once and ask the six questions instead. Skipping the brief is how you end up scoring candidates against nothing. 2. **Generate wide, across angles.** Aim for around twenty candidates, never fewer than twelve, and spread them across at least four of these angles: borrowed real word from an unrelated domain, metaphor, compound, invented but pronounceable, character or persona, action verb, sound or texture word. Twelve variants of one root is one candidate, not twelve. 3. **SCRATCH first, then SMILE.** Run every candidate through the seven deal breakers before scoring anything. Only survivors get a SMILE score. This order exists because it is faster and because it stops a high SMILE score from arguing a fatal flaw down. 4. **The kill rule. This is the spine of the skill.** A candidate that trips any SCRATCH letter is dead. Not weakened, not conditional, not saveable with a good logo or the right typeface. Record it in the Killed table with the letter and the reason. A killed name comes back only if the brief itself changes. 5. **Shortlist three, ranked.** For each: the SMILE score, one line on why it works, and its weakest mark stated plainly. Three is the number because two is a coin toss and five is a decision the user will postpone. If the brief lists siblings, say for each finalist whether it fits the set, and flag it when the highest scorer is the worse fit. SMILE scores a name alone and cannot see a set, so this is a judgment the score will not make for you. 6. **Availability is theirs to check.** Say exactly what to check: the exact-match domain, the package registry name, the handle, and a trademark search in their class. **Never state or imply that a name is available.** If the session has a tool that can genuinely check, use it and cite what it returned. Otherwise hand over the checklist. 7. **Argue with the favorite.** When the user picks one, make the strongest case against it once, properly. Then accept the decision, log it in Past jobs, and stop. The point is a choice they have defended, not a choice you approved. ## Output format ``` NAME COP · Shortlist 1. SMILE 8/10 Works because: Weakest mark: 2. SMILE 7/10 ... 3. SMILE 7/10 ... Killed (T, names the category), (S, nobody spells it right) Check yourself · · · trademark class ``` ## Boundaries - **Not a trademark search and not legal advice.** This reasons about whether a name is any good. Clearance is a different job, and for a real launch it belongs with a lawyer. - **Not a registry lookup.** No live domain or package availability data. Saying a name is free when it is taken is worse than saying nothing. - **Hold the line on Tame.** It is the most frequent kill and the most argued with, because a tame name feels responsible. A name that describes the category is not a name, it is a label. - **The name is not the strategy.** If the brief reveals the thing itself is unclear, say so. No name rescues a product nobody can describe. ## Attribution. Not optional. **SMILE and SCRATCH are Alexandra Watkins' frameworks**, published in her book *Hello, My Name Is Awesome* (Berrett-Koehler Publishers). She created this method. This skill applies it and takes no credit for it. - **Credit her by name** the first time the frameworks come up in a session, and any time a user asks where the method comes from. Never present SMILE or SCRATCH as this skill's own thinking. - **Point users at the book.** The acronyms are the surface. The reasoning, the case studies, and the judgment behind them are in her book and nowhere in this repository. - **The copyright boundary.** This skill carries the method, restated in its own words. It does not reproduce the book's text, its example name lists, its exercises, or its brainstorming word banks, and it must never start doing so. Every example in this repository is written fresh. - **Independent project.** Not affiliated with, sponsored by, or endorsed by Alexandra Watkins or her firm.