Clone it. Claude’s ready.
One repo bootstraps every skill, the global instructions and the matchmaker. File-based — no plugins, no hooks.
Two layers, one clone.
Global skills are always on. The catalog installs per project, only where it's needed — the build stays fast and the context stays lean.
14 skills + your instructions
Copied into ~/.claude by setup.sh. Available the moment you open any project — kickoff, design, engineering, ship & sessions.
- 14 skills, 5 of them your own
- The global CLAUDE.md
- Slash commands
- Set up once
A 62-skill catalog
The matchmaker reads a project's stack and installs only the skills it needs into .claude/skills/ — so your context never carries what it isn't using.
- 62 catalog skills
- Matched to your stack
- Loaded per project
- Run /skills-suggest
Skills in every project.
Set up once by setup.sh, then available everywhere. Five are ours; nine are pulled from their original sources and patched on install.
Project kickoff
ui-ux-pro-max
externalDesign direction from a real database — 67 styles, 161 palettes, 57 font pairs.
/ui-ux-pro-maxDesign
frontend-design
externalAesthetic direction — fires automatically on UI work.
impeccable
externalDesign process: critique → audit → polish. 23 commands.
/impeccableemil-design-eng
externalMotion and the invisible details — fires on UI work.
review-animations
externalStrict motion review.
/review-animationsEngineering
systematic-debugging
externalRoot cause before any fix.
debug Xverification-before-completion
externalEvidence before claiming done.
verifysupabase
externalOfficial Supabase guidance — fires on Supabase work.
supabase-postgres-best-practices
externalPostgres performance, the official way.
Ship & sessions
ship
owncommit → push → open PR, in one shot.
/shipship-merge
ownship + CI wait + light review + squash-merge + cleanup.
/ship-mergesession-handoff
ownA clean end-of-session handoff before /clear.
wrap up sessionCatalog engine
skill-matchmaker
ownInstalls catalog skills that match a project's stack.
/skills-suggestskill-scout
ownFinds new skills, plugins and MCPs in the wider ecosystem.
The 62-skill catalog.
Per-project skills the matchmaker installs where they fit. Pick the right one for the job — filter by family.
agent-harness-construction
Design and optimize AI agent action spaces, tool definitions, and observation formatting for higher completion rates. Covers granularity rules, error recovery contracts, and context budgeting.
agent-introspection-debugging
Structured self-debugging workflow for AI agent failures. Capture failures, diagnose root causes, apply contained recovery, produce debug reports.
agentic-engineering
Operate as an agentic engineer using eval-first execution, decomposition, and cost-aware model routing. Use when AI agents perform most implementation work and humans enforce quality controls.
claude-api
Build Claude API applications in Python and TypeScript. Covers Messages API, streaming, tool use, vision, extended thinking, batches, prompt caching, and Agent SDK patterns.
claude-devfleet
Orchestrate multi-agent coding tasks via Claude DevFleet — plan projects, dispatch parallel agents in isolated worktrees, monitor progress, and read structured reports.
continuous-learning
Automatically extract reusable patterns from Claude Code sessions and save them as learned skills for future use.
deep-research
Multi-source research using firecrawl and exa MCPs. Searches, synthesizes findings, and delivers cited reports with source attribution.
eval-harness
Eval-driven development framework for Claude Code. Define pass/fail criteria before coding, measure reliability with pass@k metrics, and track regressions.
exa-search
Neural search via Exa MCP for web, code, companies, and people. Access current information, code examples, technical references, and business intelligence.
iterative-retrieval
Pattern for progressively refining context retrieval to solve the subagent context problem. Agents retrieve only the context they need through iterative DISPATCH-EVALUATE-REFINE-LOOP cycles.
mcp-server-patterns
Build MCP servers with Node/TypeScript SDK. Tools, resources, prompts, Zod validation, stdio vs Streamable HTTP transport options.
prompt-optimizer
Analyze raw prompts, identify intent and gaps, match ECC components (skills/commands/agents/hooks), and output a ready-to-paste optimized prompt. Advisory role only — never executes the task itself.
audit-context-building
Enables ultra-granular, line-by-line code analysis to build deep architectural context before vulnerability or bug finding. Applies First Principles, 5 Whys, and 5 Hows at micro scale.
codeql
Scans a codebase for security vulnerabilities using CodeQL interprocedural data flow and taint tracking. Supports run-all and important-only scan modes with data extension models and SARIF output.
insecure-defaults
Detects fail-open insecure defaults — hardcoded secrets, weak auth, permissive security — that let apps run insecurely in production. Distinguishes exploitable defaults from fail-secure patterns.
sarif-parsing
Parses and processes SARIF files from static analysis tools like CodeQL, Semgrep, or other scanners. Handles filtering, deduplication, format conversion, and CI/CD integration of SARIF data.
semgrep
Run Semgrep static analysis on a codebase using parallel subagents. Supports run-all and important-only modes. Auto-detects Semgrep Pro for cross-file taint analysis when available.
supply-chain-risk-auditor
Identifies dependencies at heightened risk of exploitation or takeover. Use when assessing supply chain attack surface, evaluating dependency health, or scoping security engagements.
variant-analysis
Find similar vulnerabilities across codebases using pattern-based analysis. Use when hunting bug variants, building CodeQL/Semgrep queries, or performing systematic security audits.
bun-runtime
Bun as all-in-one JavaScript runtime, package manager, bundler, and test runner. Covers Node.js compatibility, Vercel deployment, and migration from Node.
deployment-patterns
CI/CD pipelines, Docker containerization, health checks, rollback strategies, and production readiness checklists for web apps. Use when setting up deployment infrastructure or planning releases.
docker-patterns
Docker and Compose patterns for local dev, container security, networking, and multi-service orchestration. Use when setting up containerized environments or reviewing Docker configurations.
git-workflow
Git workflow patterns including branching strategies, commit conventions, merge vs rebase, conflict resolution, and collaborative development best practices for teams of all sizes.
benchmark
Measures performance baselines, detects regressions before/after PRs, and compares stack alternatives. Tracks Core Web Vitals, bundle sizes, API latency, build times, and performance trends.
browser-qa
Automates visual testing and UI interaction verification using browser automation after deploying features. Performs smoke tests, interaction tests, visual regression, and accessibility audits.
code-review
Code review automation for TypeScript, JavaScript, Python, Go, Swift, Kotlin. Analyzes PRs for complexity and risk, checks code quality for SOLID violations and code smells, generates review reports.
e2e-testing
Playwright E2E testing patterns, Page Object Model, configuration, CI/CD integration, artifact management, and flaky test strategies for building stable, fast, and maintainable test suites.
mutation-testing
Configures mewt or muton mutation testing campaigns — scopes targets, tunes timeouts, and optimizes long-running runs. Use when setting up or tuning mutation testing for code quality assurance.
property-based-testing
Property-based testing guidance across languages and smart contracts. Generates stronger coverage than example-based tests for serialization, parsing, validation, and algorithmic code.
second-opinion
Runs external LLM code reviews (OpenAI Codex or Google Gemini CLI) on uncommitted changes, branch diffs, or specific commits. Provides independent review powered by a separate model.
tdd-workflow
Test-driven development with 80%+ coverage including unit, integration, and E2E tests using Jest/Vitest, Playwright, and mocking patterns.
html-ppt-pitch-deck
Investor-ready 10-slide HTML pitch deck — white + blue→purple gradient hero, big numbers, traction bar chart, ask page. Use for seed-round pitches and VC fundraising.
html-ppt-product-launch
Launch keynote HTML deck — dark hero + light content, warm orange→peach accent, feature cards, pricing tiers, CTA. Use for product announcements and feature reveals.
html-ppt-tech-sharing
Conference/tech-talk HTML deck — GitHub-dark, JetBrains Mono, terminal code blocks, agenda + Q&A pages. Use for engineering talks and technical presentations.
html-ppt-zhangzara-monochrome
Monochrome HTML deck — ivory ledger paper, all-black type, Lora serif headlines. Use for white papers, user research, academic briefs, policy documents.
kami-deck
Produce a print-grade slide deck in the kami design system—warm parchment, serif typography, ink-blue accent, magazine-style swipe navigation.
blog-post
Long-form article/blog post — masthead, hero image, article body with pull quotes, author byline, related posts. Use for blog articles, essays, case studies, newsletters.
dashboard
Admin/analytics dashboard in single HTML file — left sidebar nav, top bar with search, KPI cards, charts. Use when brief asks for dashboard, admin panel, analytics, or control panel.
design-brief
Parse natural-language design briefs into structured specs using I-Lang protocol—eliminate ambiguity with explicit palette, typography, layout, mood, and density.
email-marketing
Brand product-launch email — masthead, hero block, headline lockup, body copy, CTA, specs grid. Pure HTML email layout (600px column). Use for email templates and newsletter blasts.
invoice
Printable invoice page — sender/recipient blocks, line items table, tax breakdown, totals, payment instructions. Use for billing documents and invoices.
live-dashboard
Notion-style team dashboard — KPIs, 7-day sparkline, activity feed, linked-database table. Wired to Notion via connectors. Falls back to mock data offline.
pricing-page
Standalone pricing page — header, plan tiers with pricing, feature comparison table, FAQ. Use for pricing pages, subscription plans, tier comparisons.
saas-landing
Single-page SaaS landing — hero, features, social proof, pricing, footer CTA. Respects active DESIGN.md tokens. Use for product landing pages and marketing sites.
social-carousel
Three-card social carousel — 1080×1080 cinematic squares with connected headlines, brand marks, loop affordance. Use for Instagram/LinkedIn carousel posts and social series.
tailwindcss-v4
Tailwind CSS v4 — CSS-first config (no tailwind.config.js by default), @theme directive, new @utility / @variant, breaking changes from v3, and the @tailwindcss/postcss vs Vite plugin split.
video-shortform
Short-form video generation — 3-10 second clips for product reveals, motion teasers, ambient loops. Works with Seedance 2, Kling 3/4, Veo 3. Outputs MP4 to project folder.
web-prototype
General-purpose desktop web prototype — single self-contained HTML file built from seed template + section layouts. Default for landing/marketing/docs pages.
wireframe-sketch
Generate hand-drawn wireframe sketches with graph-paper background, marker tones, and lo-fi annotation—perfect for early-stage design exploration.
api-design
REST API design patterns including resource naming, HTTP methods, status codes, pagination, filtering, error responses, versioning, rate limiting, and authentication.
backend-patterns
Backend architecture patterns including repository/service layers, database optimization, caching, error handling, middleware, authentication, rate limiting, and job queues.
database-migrations
Zero-downtime schema migrations, rollbacks, and data migrations across PostgreSQL, MySQL, and common ORMs (Prisma, Drizzle, Django, TypeORM). Use when planning or implementing database schema changes.
drizzle-neon-typescript
Drizzle ORM on Neon serverless Postgres in TypeScript — schema definition, type-safe queries, migrations via drizzle-kit, connection pooling, and the serverless vs edge driver split.
postgres-patterns
PostgreSQL patterns for query optimization, schema design, indexing, and security. Quick reference for index types, data types, and anti-pattern detection. Based on Supabase best practices.
supabase-typescript
Supabase patterns in TypeScript — typed client setup, RLS-friendly queries, auth flows, realtime subscriptions, and the SSR/server-action handoff.
vercel-blob-storage
Vercel Blob — uploading from server actions or client, the put/del/list/head API, public vs private blobs, signed client uploads for large files, and the path/pathname conventions.
custom-jwt-auth-jose
Custom JWT auth in Next.js with jose + bcryptjs — sign/verify tokens, httpOnly cookies, middleware-friendly verification (Edge-compatible), password hashing, and the server-action login flow.
frontend-patterns
React, Next.js, state management, performance optimization, and UI best practices with patterns for composition, hooks, caching, and accessibility.
nextjs-app-router
Next.js App Router patterns — server components, server actions, route handlers, layouts, streaming, and the boundaries between server and client code.
nextjs-turbopack
Next.js 16+ with Turbopack for incremental bundling, faster dev startup, HMR, file-system caching, and production bundle optimization.
pwa-deployment
PWA essentials — manifest, service worker registration, install prompt, iOS quirks, and deploying so the app actually installs.
react-19-patterns
React 19 patterns — Actions, `useFormStatus`/`useFormState`/`useActionState`, `useOptimistic`, the `use()` hook, ref-as-prop, document metadata, and what's different from React 18.
File-based. No plugins, no hooks.
Rebuilt from an audit of 38 sessions. The old build ran 8 plugins and automatic hooks; the latency and context bloat had to go.
Files, not plugins
Skills are plain files in ~/.claude/skills/. It works on the desktop app — the claude CLI isn't required.
Zero hooks
Latency was the #1 complaint. Nothing fires automatically on every edit; enforcement lives in the instructions.
Externals from source
The 9 external skills are pulled from their upstreams and patched on install — never vendored, so updates are one command.
Audited, not accumulated
An audit of 38 sessions cut dead connectors and heavy skills. Every global skill has to earn its start-up cost.
The full reasoning — what was cut and why — lives in DECISIONS.md.
Send it to Claude.
Paste a line and Claude does the rest — the whole build, just the skills, or just the instructions.
Everything
Skills + instructions + slash commands. The full bootstrap.
Clone https://github.com/itsjustiago/skillsbase and run `bash setup.sh` — install the 14 global skills, the global CLAUDE.md and the slash commands, then tell me what changed and remind me to restart Claude Code.Skills only
Just the global skills. Leaves your instructions untouched.
From github.com/itsjustiago/skillsbase, install only the global skills (`global-skills/` → `~/.claude/skills/`). Don't touch my CLAUDE.md or settings. List what you added.Instructions only
Just the global CLAUDE.md, with a backup of your current one.
Install only the global CLAUDE.md from skillsbase (`setup/CLAUDE.md` → `~/.claude/CLAUDE.md`), backing up my existing one first. No skills.Prefer the shell?
git clone https://github.com/itsjustiago/skillsbase && cd skillsbase && bash setup.shRequires git + node on PATH. The claude CLI isn’t required — this works on the desktop app. Restart Claude Code after installing.
Match this machine to the repo
Reconcile an existing machine — dry run first.
Make this machine match skillsbase exactly: run `bash sync.sh` (dry run), show me the diff, and apply it only if I say OK.Add per-project skills
Let the matchmaker pick catalog skills for a project.
Run /skills-suggest — pick the catalog skills that match this project's stack and install them into .claude/skills/.Everything comes from one public repo — github.com/itsjustiago/skillsbase.