A county engineer wants a personal brainstorming skill that teammates should never inherit. Where should it live?

When a county engineer wants a personal brainstorming skill that teammates must never inherit, the correct placement is user-scoped skills with a distinct name so only that engineer loads it. Claude Code separates project-shared skills from personal user-scoped skills precisely so individual workflows stay local while team workflows remain in the repository for every clone.

User-scoped skills work because they live outside the shared project tree that clones and CI pick up. A distinct name avoids colliding with team skill names and makes the personal brainstorming helper discoverable only in that engineer's environment. The rest of the county staff continue with project skills and CLAUDE.md standards without inheriting one person's brainstorming prompts or private scaffolding habits.

Placing the skill in project .claude/skills/ fails by concept because project skills are inherited by every clone of the municipal repository, so teammates would automatically receive the personal brainstorming skill. Putting it in root CLAUDE.md fails because CLAUDE.md is always-on shared project instruction, not an on-demand personal skill, and it would load for all staff on every session. A public wiki page that CI injects into every pull-request review fails because that path publishes and re-injects the content into shared review surfaces rather than keeping it private to one engineer.

Exam caveat: skill location (user versus project) is independent of whether content is a skill versus CLAUDE.md; personal exclusivity requires user scope, not merely a clever skill name in the project tree. Operational check: create the brainstorming skill under user scope, clone the county repo on a second machine without that user skill, and verify the personal skill does not appear for teammates.