A curated collection of skills that supercharge Claude Code — turning it into a domain expert on demand.
Skills are structured knowledge packs that Claude Code loads contextually. When you invoke a skill, Claude gains deep expertise in that domain: the right frameworks, best practices, proven patterns, and production-ready code templates. No hallucination, no outdated APIs — just precise, battle-tested guidance.
Requires Bun v1.0+ for gstack browser skills. Install with
curl -fsSL https://bun.sh/install | bash
# Clone into your Claude Code skills directory
git clone https://github.com/almeidamarcell/claude-code-skills.git ~/.claude/skills
# Run setup (required — links impeccable skills, syncs settings, builds gstack)
~/.claude/skills/setup.shWhy is
setup.shrequired? Claude Code only discovers skills one level deep (~/.claude/skills/*/SKILL.md). The 21 Impeccable design skills and 17 SEO skills live underimpeccable/andclaude-seo/(two levels deep) and won't be visible without the symlinks thatsetup.shcreates. Always runsetup.shafter cloning or pulling updates.
cd ~/.claude/skills && git pull && ./setup.shThis syncs skills, recreates impeccable and SEO symlinks, and updates plugin settings (settings.json) across all your devices.
# Copy just the skill you need (top-level skills only — impeccable skills require setup.sh)
mkdir -p ~/.claude/skills
cp -r path/to/skill ~/.claude/skills/For teams where each member may have their own settings.json or doesn't need every skill.
git clone https://github.com/almeidamarcell/claude-code-skills.git ~/.claude/skills
~/.claude/skills/setup.shThis gives you everything: all skills (including impeccable and SEO), shared settings, and gstack browser daemon.
git clone https://github.com/almeidamarcell/claude-code-skills.git ~/.claude/skills
~/.claude/skills/setup.sh --skills-onlyInstalls all skills and impeccable/SEO symlinks but doesn't overwrite your existing settings.json.
git clone https://github.com/almeidamarcell/claude-code-skills.git ~/.claude/skills
~/.claude/skills/setup.sh --skills-only --no-gstackInstalls all skills (including impeccable and SEO) without building the gstack browser daemon. No Bun dependency required.
If your team already has a skills repo or wants to vendor specific skills:
# Copy individual skills into your team's skills directory
cp -r tdd/ shaping/ safe-deploy/ /path/to/your-team-skills/
# For impeccable skills, copy each sub-skill directly (not the parent directory)
for skill in impeccable/*/; do cp -r "$skill" /path/to/your-team-skills/; done
# For SEO skills, copy each sub-skill directly (not the parent directory)
for skill in claude-seo/*/; do cp -r "$skill" /path/to/your-team-skills/; done
# For gstack skills, copy the whole gstack/ directory (it has shared dependencies)
cp -r gstack/ /path/to/your-team-skills/
cd /path/to/your-team-skills/gstack && ./setup- Clone the repo —
git clone https://github.com/almeidamarcell/claude-code-skills.git ~/.claude/skills - Run setup (always required for impeccable and SEO skills):
- Full install (solo/opinionated):
./setup.sh - Skills only (keep your settings):
./setup.sh --skills-only - Lightweight (no browser):
./setup.sh --skills-only --no-gstack
- Full install (solo/opinionated):
- Restart Claude Code for skills to take effect
- Verify impeccable skills — run
/polishor/critiquein Claude Code to confirm they work - Update anytime —
cd ~/.claude/skills && git pull && ./setup.sh [same flags]
Skills activate automatically when Claude Code detects a matching context, or invoke them explicitly:
/tdd # Test-driven development workflow
/shaping # Shape a solution collaboratively
/animejs # Generate anime.js v4 animations
/changelog # Auto-generate changelogs from commits
/clearshot # Analyze a UI screenshot
# gstack skills
/browse # Real Chromium browser automation
/qa # QA testing with bug fixes
/review # Staff engineer code review
/ship # Release engineer workflow
/office-hours # YC-style product review
| Skill | Description |
|---|---|
| tdd | Strict test-driven development workflow. Red-green-refactor with zero exceptions |
| mcp-builder | Build high-quality MCP (Model Context Protocol) servers in TypeScript or Python |
| web-artifacts-builder | Create multi-component React + Tailwind + shadcn/ui artifacts for claude.ai |
| animejs | Generate production-ready anime.js v4 animations — DOM, SVG, scroll-triggered, timelines, stagger, draggable |
| cheap-model-testing | Always use the cheapest LLM model during dev/testing — save costs, upgrade for prod |
| changelog | Auto-generate user-facing changelogs from git commit history |
| safe-deploy | Pre-deploy safety checks — verifies branch ancestry before shipping |
| Skill | Description |
|---|---|
| shaping | Collaborative problem definition and solution design using the Shape Up methodology |
| breadboarding | Map systems into affordance tables and wiring diagrams |
| clearshot | Structured screenshot analysis for UI implementation and critique — 5x5 spatial grid, element inventory, design system extraction |
| anthropic-brand-guideline | Apply Anthropic's official brand colors, typography, and design standards |
A collection of design-focused skills from pbakaus/impeccable for UI/UX refinement. Each targets a specific aspect of interface quality.
Important: Impeccable skills live under
impeccable/but Claude Code only discovers skills one level deep (~/.claude/skills/*/SKILL.md). You must runsetup.shto create symlinks at the top level (e.g.,~/.claude/skills/polish→impeccable/polish). Without this step, impeccable skills will not be available in Claude Code. If you cloned without runningsetup.sh, run it now.
| Skill | Description |
|---|---|
| adapt | Adapt designs across screen sizes, devices, and contexts |
| animate | Enhance features with purposeful animations and micro-interactions |
| arrange | Improve layout, spacing, and visual rhythm |
| audit | Comprehensive audit of accessibility, performance, and interface quality |
| bolder | Amplify safe or boring designs to be more visually stimulating |
| clarify | Improve UX copy, error messages, microcopy, labels, and instructions |
| colorize | Add strategic color to monochromatic interfaces |
| critique | Evaluate design effectiveness from a UX perspective |
| delight | Add moments of joy and personality to interfaces |
| distill | Strip designs to their essence by removing unnecessary complexity |
| extract | Extract reusable components, design tokens, and patterns |
| frontend-design | Create distinctive, production-grade frontend interfaces |
| harden | Improve resilience — error handling, i18n, text overflow, edge cases |
| normalize | Normalize design to match your design system |
| onboard | Design onboarding flows, empty states, and first-time user experiences |
| optimize | Improve performance — loading, rendering, animations, image optimization |
| overdrive | Push interfaces past conventional limits — shaders, physics, 60fps animations |
| polish | Final quality pass — alignment, spacing, consistency, detail fixes |
| quieter | Tone down overly bold or aggressive designs |
| teach-impeccable | One-time setup — gathers design context for your project |
| typeset | Improve typography — font choices, hierarchy, sizing, weight, readability |
A comprehensive SEO toolkit from AgriciDaniel/claude-seo covering technical audits, content quality, local SEO, schema markup, and more.
Important: SEO skills live under
claude-seo/but Claude Code only discovers skills one level deep (~/.claude/skills/*/SKILL.md). You must runsetup.shto create symlinks at the top level (e.g.,~/.claude/skills/seo→claude-seo/seo). Without this step, SEO skills will not be available in Claude Code.
| Skill | Description |
|---|---|
| seo | Comprehensive SEO analysis — full site audits, E-E-A-T, Core Web Vitals, schema, GEO |
| seo-audit | Full website technical audits with parallel subagent delegation |
| seo-page | Deep single-page SEO analysis |
| seo-technical | Technical SEO audits — crawlability, indexability, security, Core Web Vitals |
| seo-content | Content quality and E-E-A-T analysis with AI citation readiness |
| seo-schema | Schema.org structured data detection, validation, and generation |
| seo-sitemap | XML sitemap analysis and generation |
| seo-images | Image optimization — alt text, file sizes, formats, lazy loading |
| seo-image-gen | AI image generation for SEO assets (OG images, hero images, infographics) |
| seo-plan | Strategic SEO planning — SaaS, e-commerce, local, publisher, agency templates |
| seo-local | Local SEO — Google Business Profile, NAP consistency, citations |
| seo-maps | Maps intelligence — geo-grid rank tracking, GBP auditing, review analysis |
| seo-geo | Generative Engine Optimization — AI Overviews, ChatGPT search, Perplexity |
| seo-hreflang | Hreflang and international SEO validation |
| seo-competitor-pages | Competitor comparison and alternatives page generation |
| seo-programmatic | Programmatic SEO for pages generated at scale |
| seo-dataforseo | Live SEO data via DataForSEO — SERP analysis, keyword research, backlinks |
gstack by Garry Tan — 27 skills that structure development as a complete sprint cycle: Think → Plan → Build → Review → Test → Ship → Reflect. Includes a persistent Chromium browser daemon for real QA testing.
Requires Bun v1.0+ — gstack compiles a browser daemon for QA/browsing skills. Built automatically by
setup.sh.
| Skill | Description |
|---|---|
| office-hours | YC Office Hours — six forcing questions that reframe your product before coding |
| plan-ceo-review | CEO/founder-mode plan review — rethink the problem, find the 10-star product |
| plan-eng-review | Eng manager review — lock in architecture, data flow, edge cases, tests |
| plan-design-review | Senior designer review — rate design dimensions 0-10, detect AI slop |
| design-consultation | Design partner — build a complete design system from scratch |
| autoplan | Auto-review pipeline — runs CEO → design → eng review automatically |
| Skill | Description |
|---|---|
| review | Staff engineer code review — find production bugs that pass CI |
| qa | QA lead — test app, find bugs, fix with atomic commits, re-verify |
| qa-only | QA reporter — same as /qa but report-only, no code changes |
| design-review | Designer who codes — design audit + fixes with before/after screenshots |
| investigate | Systematic root-cause debugging — traces data flow, tests hypotheses |
| codex | Second opinion — independent code review from OpenAI Codex CLI |
| Skill | Description |
|---|---|
| ship | Release engineer — sync main, run tests, audit coverage, push, open PR |
| land-and-deploy | Merge PR, wait for CI, deploy, verify production health |
| canary | Post-deploy canary monitoring — watches for errors and regressions |
| benchmark | Performance engineer — baseline Core Web Vitals, compare before/after |
| document-release | Technical writer — update docs to match shipped code |
| retro | Weekly engineering retrospective with per-person breakdowns |
| Skill | Description |
|---|---|
| browse | Real Chromium browser automation — real clicks, real screenshots, ~100ms per command |
| setup-browser-cookies | Import cookies from real browser for authenticated pages |
| Skill | Description |
|---|---|
| cso | Chief Security Officer — OWASP Top 10 + STRIDE threat model |
| careful | Warns before destructive commands (rm -rf, DROP TABLE, force-push) |
| freeze | Restrict file edits to one directory |
| guard | Full safety — /careful + /freeze combined |
| unfreeze | Remove the /freeze boundary |
| Skill | Description |
|---|---|
| setup-deploy | One-time deploy configuration for /land-and-deploy |
| gstack-upgrade | Self-updater — upgrade gstack to latest |
25 marketing-focused skills covering A/B testing, analytics, copywriting, CRO, email sequences, SEO, pricing, and more. These have not been validated yet — use at your own risk.
See mkt-skills(not-tested)/ for the full list.
Two project-level slash commands from almeidamarcell/claude-laravel-skills for Laravel apps. These are slash commands (not globally-installed skills) — copy them into your project's .claude/commands/ directory.
| Command | Description |
|---|---|
| qa | Functional QA via Chrome DevTools MCP — tests flows as a real user, finds bugs, and fixes them in the same session |
| audit | Full pentest — static tools (composer audit + Enlightn) plus active browser exploration, report-only with Laravel-specific remediation |
Install per project:
cp claude-laravel-skills/qa.md claude-laravel-skills/audit.md your-project/.claude/commands/See claude-laravel-skills/README.md for usage, configuration, and adapting to non-Laravel stacks.
Each skill is a directory containing a SKILL.md file with optional reference materials:
skill-name/
SKILL.md # Main skill file (YAML frontmatter + markdown)
references/ # Optional supporting docs
patterns.md
frameworks.md
SKILL.md format:
---
name: skill-name
description: When and how this skill should be activated...
---
# Skill Title
Detailed guidance, API references, code examples,
best practices, and production-ready patterns.- Create a directory under
~/.claude/skills/your-skill-name/ - Add a
SKILL.mdwith YAML frontmatter (name,description) - Write comprehensive guidance — Claude performs best with detailed, opinionated instructions
- Add a
references/directory for extended lookup tables or pattern libraries - The
descriptionfield doubles as the activation trigger — include phrases users might say
Contributions welcome! Whether it's a new skill, improvements to existing ones, or bug fixes:
- Fork the repo
- Create your branch (
git checkout -b feat/awesome-skill) - Commit your changes
- Push and open a PR
This project is open source under the Apache License 2.0.
