Get naming suggestions
Eight name candidates for one thing, each tagged with its convention and the collision risk it carries — plus which one to pick.
Last updated
Fill in the blanks
Assembled in your browser — nothing you type is stored or sent anywhere.
Your prompt
Model-agnostic — works in Claude, ChatGPT, Gemini or any AI chat.
Why this prompt works
- →Fixing the count at exactly 8 forces real breadth — enough options to compare naming styles against each other, instead of settling for the first name a model reaches for.
- →Naming the convention behind each candidate is what lets you keep a codebase self-consistent, rather than quietly introducing a new casing style one function at a time.
- →Calling out collision risk by name catches the specific way generated names go wrong: one that reads fine alone but is a near-duplicate of an identifier already in use.
Part by part
- Give exactly 8 candidates
- A specific number rather than "some". It forces the model past the obvious two or three into the range where genuinely different framings appear.
- the naming convention it follows
- Tagging each candidate makes the list comparable. Without it you get eight names in three conventions and no way to see that.
- one line on any collision or confusion risk
- The part that catches near-homonyms and reserved words, which is where a name actually goes wrong rather than where it merely reads badly.
- say which one you'd pick and why
- Forces a commitment. A list of eight equal options moves the decision back to you having done none of the thinking.
- instead of inventing a codebase you haven't seen
- Names the specific hallucination this prompt invites — confident claims about collisions with code the model has never been shown.
Common mistakes
What should I call this function?Returns one or two names with no reasoning, and no indication of what convention they follow or what they might collide with. The answer is unimprovable because nothing about it is explained.
Give me some name ideas.Vague quantity produces three, usually variations on the same word. The specific count is what pushes past the first idea into alternatives worth comparing.
Suggest names that follow our conventions.The model does not know your conventions unless you paste them. Asked this way it invents a plausible convention and grades against that, which reads as authoritative and is not.
Questions
Why eight rather than three?
Because the first three are the obvious ones. Requiring eight forces the model into different framings — domain terms, abbreviations, verb-first forms — and the useful candidate is usually somewhere past the fifth.
How much context is worth pasting?
Enough for the collision check to mean anything: the surrounding module's existing names, or the vocabulary your domain already uses. Without it, that section of the answer is guesswork dressed as analysis.
Should I take the recommendation?
Use it as a tiebreak rather than a verdict. The model can rank clarity and convention fit, and it cannot know which term your team already argues about or which one a customer would recognise.
Does this work for naming things other than code?
Yes, and the convention field becomes less useful. For a product or feature name the collision check matters more, since the risk is a trademark or an existing product rather than a reserved word.