| name | emmi |
| version | 0.3.1 |
| description | Emmi opinionated skill kit: iterative product planning (MVP to production), UI/UX with block library, hard QA (ferdig-ferdig), testing and CI/CD, Railway-first deployment, Next.js+tRPC+Prisma+Better Auth+Tailwind, Expo+NativeWind, Turborepo, performance work, architecture, product and team management, browse, and analytics. Use when the user wants Emmi workflows, sub-skill by name, or "the kit". Read sub-skills from this repo root. |
Emmi
Communication style
- Never use emojis in any output, code comments, docs, or generated content unless the user explicitly requests them.
- Save tokens: default to minimal prose. Skip step-by-step narration, “I will…”, and play-by-play of edits. Let diffs and code speak; only add text when the user asked a question, you must warn (risk, ambiguity), or a one-line summary materially helps review.
Update check (optional)
./bin/emmi-update-check 2>/dev/null || ~/.claude/skills/emmi/bin/emmi-update-check 2>/dev/null || true
If output includes UPGRADE_AVAILABLE, tell the user to git pull in the Emmi repo (or reinstall via ./bin/setup).
Kit root
When this file lives at ~/.claude/skills/emmi/SKILL.md, all paths below are relative to ~/.claude/skills/emmi/. In a git clone, they are relative to the repository root.
Governance
rules/: Patterns to avoid. Do not add or edit entries without asking the user. Propose candidates when you see repeated mistakes.
evaluators/: Signals of good solutions. Do not add or edit without asking. Propose metrics when they would sharpen decisions.
Read existing entries at session start when doing substantive product or UI work.
Sub-skills (read the matching SKILL.md)
| Sub-skill | When |
|---|
| iterative-process | New features, new projects, nothing → MVP → production plan, feedback rounds |
| ui-ux | UI from blocks + ecommerce / marketing / application rules |
| ferdig-ferdig | Ship-quality QA, Playwright (web) or Maestro (mobile), adversarial use |
| test | Unit through smoke tests, CI/CD, regression prevention |
| deployment | Railway-first deploy, env, rollbacks |
| web-dev | Next.js, Better Auth, Prisma, Tailwind, tRPC |
| mobile-dev | Expo, NativeWind |
| monorepo-dev | Turborepo combining web and mobile |
| efficiency-dev | Load time, efficiency, scaling |
| architect-dev | System and data-flow planning |
| product-manager | Product scope, priorities, outcomes |
| team-manager | Tasks, process, GitHub |
| browse | Headless browser testing (see skill; optional tooling there) |
| analytics | PostHog and related instrumentation |
SEO (external skill — always prefer)
For any SEO-related task, defer to the dedicated SEO skill instead of improvising. These are not Emmi sub-skills — they live externally and are invoked directly.
If the skill is not installed, install it first:
curl -fsSL https://raw.githubusercontent.com/AgriciDaniel/claude-seo/main/install.sh | bash
Source: https://github.com/AgriciDaniel/claude-seo
| Intent | Invoke |
|---|
| Full site audit | /seo or /seo-audit |
| Single page analysis | /seo-page |
| Technical SEO (crawlability, Core Web Vitals, indexability) | /seo-technical |
| Schema / structured data | /seo-schema |
| Content quality, E-E-A-T | /seo-content |
| Sitemap | /seo-sitemap |
| Backlinks | /seo-backlinks |
| Local SEO, Google Business Profile | /seo-local |
| Maps intelligence, geo-grid ranking | /seo-maps |
| GEO / AI Overviews / ChatGPT / Perplexity visibility | /seo-geo |
| Google APIs (Search Console, CrUX, GA4) | /seo-google |
| DataForSEO data (SERP, keywords, live data) | /seo-dataforseo |
| Image SEO and OG images | /seo-images |
| Hreflang / international SEO | /seo-hreflang |
| Programmatic SEO planning | /seo-programmatic |
| Competitor pages | /seo-competitor-pages |
| Performance (Core Web Vitals lab data) | /seo-performance |
| SEO strategy and planning | /seo-plan |
UI blocks index
Regenerate block indexes after library changes:
python3 ui-blocks/scripts/regenerate-catalog.py
Project routing
Copy or merge CLAUDE.md into an application repository so agents pick the right Emmi sub-skill.