consistent-naminglisted
Install: claude install-skill thatdudealso/consistent-naming
# Consistent naming
Stop random names. Every new name must look like it already belonged in the
**target** tree - files included.
## Hard rule
Before proposing or writing any new name:
1. Pick the **naming scope** ([references/scope.md](references/scope.md)):
siblings → package → git root → sibling projects on disk → house defaults.
2. Run `scripts/survey-names.sh` on that scope (and `--siblings` for new projects).
3. Read [references/rules.md](references/rules.md) for the name kind.
4. Propose **one primary name** plus at most **two alternates**, each justified
by concrete examples from the survey.
5. If patterns conflict, prefer the **local neighborhood**, then one sentence
explaining why.
Do not ship cute, meme, or one-off names that ignore neighbors. Do not reuse
repo A's style inside repo B without surveying B.
## Workflow
1. **Classify** the name kind ([references/kinds.md](references/kinds.md)):
project, package, directory, **file**, **system/config file**, component,
test, env, branch, skill, cli, api/route.
2. **Survey** the owning path (not a random workspace root):
```bash
scripts/survey-names.sh [path]
# new project / product stem also:
scripts/survey-names.sh --siblings [parent-of-projects]
```
3. **Match case and separators** for that kind in that folder
(kebab dirs, PascalCase components, SCREAMING_SNAKE env, fixed tool filenames).
4. **Match vocabulary** already in that tree. Reuse stems; avoid synonym forks
(`auth