Personal agent skills for Claude Code, installable with the
skills CLI.
npx skills add joydeeproni/skillsor a single skill:
npx skills add joydeeproni/skills --skill this-doesnt-look-goodAdd --global to install at user level (available in every project) instead of just the
current one. npx skills add joydeeproni/skills --list shows what's in the repo without
installing.
Turns vague design/UX feedback into actionable direction. Triggers on any non-specific complaint — "this looks bad", "make it prettier", "improve the ux" — and stops before touching code to interrogate the feedback through seven lenses (hesitation, expectation gaps, emotion, what's missing, what's assumed, how it looks, how it feels), reframe the problem (add/remove a constraint, invert it), synthesize concrete problem→fix pairs, and confirm before implementing. Skips itself when the instruction is already specific ("change font to 14px").
Grows a single long-form, editorial UX case study in Notion from a repo's commit history. As the product's UX is redesigned commit-by-commit, it extends one evolving narrative rather than a changelog — and when a change's why isn't clear from the diff, it asks the author for the design rationale before writing it up.
Project-agnostic: the Notion target, product name, and watched paths come from a per-repo
.claude/casestudy/config.json. See the skill's Setup section (and skills/ux-case-study/setup/)
for the one-time install of the commit-capture git hook and the optional Claude Code nudge hook.