Pedro Nauck's Skills

A curated collection of 131 agent skills for Claude Code and compatible AI coding assistants — 38 original (⭐️), 29 hand-picked (💎), 18 marketing & business (📣), plus 46 community skills. Each skill provides domain-specific knowledge, best practices, and guided workflows that enhance an agent's ability to perform specialized tasks.

Installation

Quick install (recommended)

Install this repository into your agent skills directory with the skills CLI:

npx skills add https://github.com/pedronauck/skills

Install a single bucket

Use the owner/repo/<subpath> shorthand to install only one bucket:

# Only the original skills
npx skills add pedronauck/skills/skills/mine

# Only the curated skills
npx skills add pedronauck/skills/skills/curated

You can also pin a specific skill with --skill:

npx skills add pedronauck/skills/skills/mine --skill react

Manual install

Copy or symlink the skills you need into your Claude Code configuration:

# Copy a single skill
cp -r skills/mine/react ~/.claude/skills/react

# Or symlink an entire bucket
ln -s $(pwd)/skills/mine ~/.claude/skills/mine

Skills are organized into four top-level buckets:

• skills/mine/ — 38 original skills authored in this repository (⭐️)
• skills/curated/ — 29 hand-picked community skills (💎)
• skills/marketing/ — 18 marketing, business, and writing skills (📣)
• skills/community/ — 46 broader community skills

Usage

Skills are automatically picked up by Claude Code when placed in the ~/.claude/skills/ directory. The agent matches tasks to relevant skills based on the description field in each SKILL.md frontmatter.

What are Skills?

Skills are structured instruction sets that give AI agents deep expertise in specific domains. Each skill lives in its own directory under skills/ and contains a SKILL.md file with metadata, procedures, and reference material. Skills follow the agentskills.io specification.

Skill Catalog

⭐️ = original skill authored in this repository  ·  💎 = hand-picked community skill  ·  📣 = marketing & business skill

Mine ⭐️

Original skills authored in this repository.

• agent-exploration — Dispatch scoped-write explorer subagents in parallel through the current harness's native subagent facility for multi-area research — each slice writes one seven-section analysis file, and the parent synthesizes a summary
• app-renderer-systems — Domain feature systems organized under a systems/ directory
• architectural-analysis — Deep architectural audit for dead code, duplication, anti-patterns, and code smells
• bubbletea — Build terminal UIs with Go and Bubbletea -- Elm architecture, Lipgloss styling, dual-pane layouts, and reusable components
• deep-review — CodeRabbit-grade deep review of a branch diff, working tree, or GitHub PR at any size (no file cap) — funnels changed files into cohorts, fans out parallel hunk-level reviewers with checkout-verified evidence discipline, merges findings mechanically, and emits a walkthrough, severity-tagged findings with committable suggestions plus AI-agent fix prompts, and a SHIP/FIX_BEFORE_SHIP/REWORK verdict — with incremental re-reviews, optional PR publishing, cross-LLM --subagent runtimes, and --spec conformance gating
• deslop — Remove AI-generated code slop from the branch diff — unnecessary comments, abnormal defensive checks, any casts, deep nesting — before claiming a task complete or opening a PR
• drizzle-safe-migrations — Production-safe Drizzle migration workflows for schema changes
• git-rebase — Git rebase operations and merge conflict resolution with clean history
• golang-master — Go 1.21–1.26 engineering doctrine for any Go codebase — a ten-rule floor (error wrapping, goroutine ownership, context-first APIs, race-clean tests) plus branch-routed references: errors (sentinel vs typed, %w chains, single-handling rule, panic policy), concurrency (channel/mutex/atomic and WaitGroup/errgroup decision tables, errgroup.SetLimit pools, spawn checklist), context (cancellation discipline, WithoutCancel detachment, value hygiene), safety (nil interface trap, append aliasing, numeric truncation, defer-in-loop), interfaces & generics (consumer-side contracts, cmp.Ordered constraints), naming & style (anti-stutter, functional options, no init()), assertion-framework-free testing (table-driven, synctest, goleak, b.Loop()), pprof-first performance methodology, version-by-version modernization tables, and module layout. Supersedes the curated golang-pro
• herdr-orchestration — Orchestrate Claude and Codex worker TUIs from a controller agent — one worker per named herdr tab, driven over the herdr socket CLI; supports plan-first delegation (Claude Code plan mode, Codex Plan mode) and native agent-status waits; controller owns assignment, state, conflict control, integration, and retirement (verified workers' tabs are closed, never left piled up); workers launch as interactive TUIs via herdr agent start
• insta-master — Plan, create, distribute, and monetize Instagram content by combining two complementary methodologies (pt-BR): Hyeser (tactical creator, 421k — Reels, virality, engagement, faceless monetization) and Rafael Kiso (mLabs founder — algorithm-as-graph, per-surface retention, internal search/SEO/AEO, consumer journey, social-media-as-a-business), distilled from 8 deep-research slices over 344 transcripts — the retention + social-signals distribution engine with per-surface thresholds (Feed 10s / Explore 11s / Reels 15s) and a diagnosis funnel, the COCA×journey content matrix (Growth / Objection / Connection / Authority), Reels craft ("it's the start, not the 3 seconds", 33% retention rule, hook→development→loop structure, lo-fi CapCut editing, A/B test reels), modern discovery (caption-as-semantic-field with the Question→Answer template, hashtag-as-SEO timeline, AEO to appear in ChatGPT/IA), profile foundations (bio-promise, subniche, @name, 0→1000 plan, converting bio link), Stories & cadence (3-5 story blocks, the frequency band over a magic number, best-time myth), creator monetization & selling (sell-without-looking-like-an-ad 80/20, 4-step DM social selling, affiliate / clips / faceless-IA / IG Shop / infoproduct), social-media-as-a-business (value/ROI pricing, 4-pillar method, media kit, UGC × influencer × brandlover), a publication checklist, plus read-only retencao-check.py + post-check.py helpers
• kb-yt-channel — Turn a YouTube channel into a Karpathy KB topic — resolves recent or full uploads, scaffolds yt-channels topics, ingests transcripts via kb ingest youtube (captions / auto / STT), and validates plus indexes the result
• no-workarounds — Enforce root-cause fixes over workarounds, hacks, and symptom patches
• qa-execution — Run real-user dogfooding sessions through public interfaces: persona-driven journey walks via browser under strict persona-fidelity guardrails (no dev shortcuts, stall-is-a-finding, evaluator framing forbidden), thematic tours, user edge probes, experiential lenses, and paper-cut hunting. Reads plans from the living docs/qa/ tree, dedups against the global bug registry before filing, applies the fix-loop governor (auto-fix only small and contained, with a red-before/green-after regression test; escalate the rest to "Decisions for a Human"), updates scenario-file verdicts, and writes an incremental per-run report — session debriefs included — created the moment the session matrix exists, with a full-suite exit gate before Final Status.
• qa-report — Plan real-user QA as living repo docs: owns the canonical docs/qa/ tree, merge-safe by construction — content-addressed ids and one file per scenario/bug/charter, so parallel branches never contend (enum-disciplined scenario tracker materialized into a gitignored state.csv view, global bug registry with the five-tier user-impact rubric, project personas, journey flowcharts, session charters, coverage taxonomy, automation backlog). Maps every user-visible change as a Mermaid journey flow before deriving scenarios (flows-before-matrix), plans persona×journey×tour session charters by cadence tier (smoke/targeted/full/sanity), and enforces "every journey walked by a persona this cycle" completeness — sessions, not test-case accumulation.
• react — React 19 under the React Compiler — the Rules of React as a build-time contract (purity, immutability, refs, globals, static components) with each rule mapped to its eslint-plugin-react-hooks v7 lint rule, a memoization decision table for when useMemo/useCallback/memo still earn their place, the full bail-out catalogue (three hard syntax errors plus the silent Todo skips) with "use no memo" bisection and DevTools/healthcheck detection, compiler setup per bundler and incremental adoption, effects (anti-pattern catalogue, useEffectEvent, useSyncExternalStore), component and composition patterns, React 19 TypeScript migration (ref as prop, React.JSX, useRef arity) with codemods, state placement plus interior-mutability libraries, React 19/19.2 APIs (use(), Actions, useOptimistic, <Activity>, metadata, cacheSignal), and Vitest/RTL v16 testing
• refactoring-analysis — Identify refactoring opportunities using Martin Fowler's code smells catalog with prioritized reports
• rust-best-practices — Unified Rust guidelines covering ownership, error handling, async/Tokio, traits, testing, performance, clippy, and documentation
• ship-pr — End-of-feature ritual: explore impact across docs/site/README, generate release notes (via pr-release when present, else inline from git log), assemble a complete PR description (with QA artifacts when detected), commit per the repo's commitlint, open the PR via gh, and optionally launch a CodeRabbit review-watch loop. Optional integrations (pr-release, skeeper, compozy, QA artifacts) auto-detect and skip cleanly when absent.
• spec-peer-review — Optional cross-LLM peer review of a spec (TechSpec/design doc/RFC/PRD) via Compozy — an independent model writes one scoped findings file (blockers/nits + READY/BLOCKED/NEEDS_REWORK) for user-directed incorporation. Project-agnostic with configurable --ide/--model/--reasoning runtime, six tech-agnostic quality markers, and auto-discovered project rules
• storybook-stories — Create, update, or refactor Storybook stories following project patterns
• tailwindcss — Tailwind CSS v4 patterns, design tokens, and tailwind-variants
• tanstack — TanStack Query, Router, and Form patterns for React — query keys, caching, mutations, prefetching, SSR/offline, file-based routes, search params, loaders, and Form validation
• tech-logos — Install official tech brand logos from the Elements registry via shadcn
• testing-boss — Comprehensive testing doctrine — Iron Laws, 12 positive patterns, 25 anti-patterns across five families, 7 mandatory gates for agents writing tests, flaky-test taxonomy with quarantine workflow, contract / property / mutation testing, and an LLM/agent eval primer (oracle ladder, LLM-as-judge calibration, agent trajectory vs outcome)
• to-prompt — Turn code, issues, or context into a handoff brief for another LLM — full context, zero prescribed solution — written from a bundled template to docs/prompts/<timestamp>_<slug>.md
• tweetsmash-api — TweetSmash REST API for fetching bookmarks, managing labels, filtering, and pagination
• typescript-advanced — Advanced type system -- generics, conditional types, mapped types, template literals
• ui-craft — Anti-AI-slop guardrails for UI/UX work — usability heuristics, accessibility floors, design-system discipline, 14 named slop patterns with severity tags, tunable design dials (VISUAL_VARIANCE/MOTION_INTENSITY/INFORMATION_DENSITY), scene-driven decisions, Product vs Brand registers, anti-defaults blocklist, 7 named UI archetypes, plus performance / motion / dark-mode reference packs and executable contrast + token-drift scripts
• vc-outreach — Runs the investor-facing raise motion (list → meeting → pipeline → close): partner-level tiered target lists with disqualification and referrer mapping, warm-intro blurbs and cold emails that convert to first calls, meeting prep and objection handling, the parallel-process pipeline (batching, momentum, follow-ups, health benchmarks), and diligence through close (data room, references, term-sheet-to-wire). Distilled from a 625-source fundraising corpus (NFX, Paul Graham, First Round, YC, Hustle Fund, real founder funnels with numbers). Pairs with vc-strategy and vc-pitch-deck
• vc-pitch-deck — Builds and tears down investor decks: maps the fundraising narrative onto a slide sequence, enforces per-slide requirements (problem, solution, traction, market, team, ask), covers proof slides, and reviews decks against funded-deck patterns and DocSend attention telemetry — with send-ahead, presented, and demo-day variants and a per-slide teardown rubric. Benchmarks carry source and year because deck norms drift. Pairs with vc-strategy and vc-outreach
• vc-strategy — Designs the fundraise before any artifact exists: fundability assessment against what VCs actually evaluate, the fundraising narrative and value proposition (positioning, why-now), round design (when to raise, how much, valuation, SAFE vs priced, dilution, milestones), stage benchmarks, term-sheet economics and negotiation, and devtools/AI-infra positioning (open source, platform risk, developer traction). Numbers carry source and vintage. Pairs with vc-pitch-deck and vc-outreach
• writing-agents-md — Author lean AGENTS.md/CLAUDE.md files as resident system-prompt context — the three-part rent test (delta / frequency / economy) per line, a scope ladder from global CLAUDE.md down to linked docs, form rules for rules that bind (no examples, no emphasis inflation), and Write / Trim / Gate branches
• writing-skills — Author, refactor, and debug agent skills: agentskills.io authoring procedure with metadata validator, the predictability doctrine (information hierarchy, pruning, leading words, descriptions, failure modes) with full glossary, and a loading-diagnosis pass for references the agent ignores (pointer-strength ladder). Merges the former skill-best-practices, writing-great-skills, and skill-load-tips
• writing-tech-post — SOTA authoring loop for engineering blog posts — 8 archetype contracts (launch / postmortem / migration / performance / tutorial / research-translation / AI-agent / security), 5-rung abstraction ladder with rung-whiplash diagnostic, 12-form evidence taxonomy with claim → artifact → reading cadence, 7-publisher voice matrix, four disclosure contracts (blameless / coordinated-disclosure / paper-link-first / what-we'd-do-differently), narrative arcs + closer taxonomy, and a read-only lint-post.py pre-publish gate
• xstate-store — @xstate/store v4 event-driven state for TypeScript — a primitive-selection table (store / store logic / atom / XState machine), store core (transitions, trigger/can, Immer), enqueue effects with the synchronous-determinism rule, Standard Schema contracts plus opt-in validateSchemas(), selectors and atoms (derived, reducer, config, async), createStoreLogic with input and named selectors, the persist / undoRedo / reset extensions, @xstate/store-react v2 hooks with scope guidance, pure-transition() testing and fromStore XState interop, and a v3 → v4 migration path (plus a Zustand concept map). Every sample typechecked against 4.2.2
• yc-apply — Drive a Y Combinator batch application end-to-end through a 10-phase workspace — captures the live YC form, profiles founders and stress-tests the idea via an embedded grill loop, runs a mandatory 5-agent parallel external-research pass on the startup, drafts every field with a buzzword scanner and a provenance-labeled accepted-answer rubric, generates founder-video bullet notes (no script), enforces a script-checked 10-check pre-submit gate, then unlocks a post-invite interview-prep simulator and reapplicant delta tracking. Built from 84 YC essays + 28 interview transcripts
• yt-master — Plan, package (title + thumbnail), script, and optimize YouTube videos by combining two complementary methodologies (pt-BR): Escola Para Youtubers (Caique — 50 transcripts; packaging-first / algorithm) and Camilo Coutinho (20-yr veteran — 100 transcripts; search SEO, sustainable production system, channel decisions), validated by real cross-channel metrics — the embrulho-primeiro production pyramid, 3 psychological thumbnail triggers (FOMO / pain+solution / objection-break) plus a data-driven post-publish thumbnail-swap, 8 metric-backed title formulas plus the 3 verbs (findable/clickable/shareable), a ≤30s hook with a named-technique library, block-based script template, the Problem-vs-Ambition payoff axis and an unblock gate (brain dump / camera fear), the 5-stage algorithm funnel (Impression → CTR → Retention → Satisfaction → Session) with myth-vs-reality and two discovery channels (internal funnel + Google search: 5-block description, rankable chapters, playlists), monetization & growth (YPP, YouTube Shopping, operational Brand Connect, sustainable production system "Fortaleza de Vídeos", community 15-min/day, dark/AI demonetization gate, copyright), channel decisions (naming, audience collision, restart/migrate), a publication checklist, and read-only ctr-baseline.py + title-check.py helpers

Curated 💎

Hand-picked community skills maintained in this repository.

• agent-browser — Automate browser interactions for testing, form filling, and data extraction
• architecture-diagram — Professional dark-themed system architecture diagrams as standalone HTML/SVG files
• autoresearch — Autonomously optimize any skill by running evals, mutating prompts, and keeping improvements
• brainstorming — Explore intent, requirements, and design through collaborative dialogue
• context7 — Retrieve up-to-date technical documentation, API references, and code examples for any library via Context7 CLI
• documentation-writer — Diátaxis-guided technical writing across tutorials, how-to guides, reference, and explanation quadrants
• effect-ts — Effect-TS code including setup, data modeling, error handling, and Context.Tag
• exa-web-search-free — Free AI-powered web, code, and company search via Exa MCP
• extreme-software-optimization — Profile-driven performance optimization with behavior proofs, opportunity scoring, and isomorphism guarantees
• firecrawl — Web scraping, search, crawling, and browser automation via the Firecrawl CLI
• game-changing-features — Find 10x product opportunities and high-leverage improvements
• hono — Hono framework development with documentation search and API reference
• impeccable — Production-grade frontend design and iteration — bold or quiet redesigns, live UI iteration, visual hierarchy, accessibility, motion, and design-token discipline backed by real working code
• lesson-learned — Extract software engineering lessons from git history and recent code changes
• mastra — Mastra framework for building AI agents and workflows
• next-best-practices — Next.js best practices -- file conventions, RSC boundaries, data patterns, async APIs, metadata, error handling, and optimization
• qmd — Search markdown knowledge bases, notes, and documentation using QMD
• ratatui-tui — Terminal UIs with ratatui v0.30.0+ -- Elm Architecture, StatefulWidget, async events
• sentry-cli — Sentry CLI for interacting with Sentry from the command line
• shadcn — Building UI components with shadcn/ui, Radix UI primitives, and design tokens
• systematic-debugging — Root-cause investigation before proposing fixes for bugs or test failures
• tui-design — Universal TUI design patterns -- layouts, color schemes, keyboard navigation, dashboards, and accessibility
• vercel-composition-patterns — React composition patterns for refactoring boolean prop proliferation
• vercel-react-best-practices — React/Next.js performance optimization from Vercel Engineering
• verification-before-completion — Run verification commands and confirm output before claiming success
• vitest — Fast unit testing with Vite -- Jest-compatible API, mocking, coverage, and fixtures
• xstate — XState v5 state machines, actors, and TanStack Query integration (for @xstate/store v4, use xstate-store)
• zod — Zod schema validation for type safety, parsing, and error handling
• zustand — Zustand state management patterns, store organization, and best practices

Marketing 📣

Marketing, sales, business, and writing skills.

• alex-hormozi-pitch — Create irresistible offers using Hormozi's $100M Offers methodology
• brand-storytelling — Craft compelling brand narratives and positioning
• content-research-writer — Writing partner for research, outlining, drafting, and refining content
• copywriting — Conversion copywriting for marketing pages, CTAs, and headlines
• fundraising — Plan and run early-stage fundraising with pitch narrative, investor pipeline, and outreach
• google-ads — Query, audit, and optimize Google Ads campaigns
• hormozi-ad-factory — Generate 150-750+ ad variations using Hormozi's combinatorial Hook x Meat x CTA framework
• humanizer — Remove signs of AI-generated writing from text
• pitch-deck — Generate professional PowerPoint pitch decks for startups
• pitch-deck-visuals — Investor pitch deck visuals with slide-by-slide framework and design rules
• pitch-gen — Generate startup pitch deck content with AI
• pptx-creator — Create professional PowerPoint presentations from outlines or data
• professional-communication — Technical communication for emails, team messaging, and meeting agendas
• promo-video — Create promotional videos using Remotion with AI voiceover and background music
• sales-methodology-implementer — Implement proven sales methodologies (MEDDIC, BANT, Sandler, Challenger, SPIN)
• startup-validator — Comprehensive startup idea validation and market analysis
• viz — Four visualization modes in one skill -- Excalidraw diagrams, Swiss Pulse PNG infographics, inline Visualizer widgets, and published HeyGenverse apps
• writing-clearly-and-concisely — Strunk's timeless rules for clearer, stronger, more professional prose

Community

Broader community skills.

• a11y-testing — Automated accessibility testing with axe-core, Playwright, and jest-axe
• adversarial-review — Spawn opposing AI model reviewers to adversarially challenge work
• ai-sdk — Vercel AI SDK for building AI-powered features
• argocd-expert — ArgoCD GitOps deployment, sync strategies, and production operations
• better-auth-best-practices — Better Auth TypeScript authentication framework integration
• better-auth-organization-best-practices — Multi-tenant organizations, RBAC, teams, members, and invitations with Better Auth's organization plugin
• building-components — Accessible, composable UI components with design tokens and documentation
• centrifugo — Centrifugo real-time messaging -- WebSocket PUB/SUB, channels, JWT auth, scaling
• cloudflare — Cloudflare platform -- Workers, Pages, storage, AI, networking, and security
• crafting-effective-readmes — Templates and guidance for writing README files matched to audience and project
• creating-spec — Comprehensive technical specs for SDK gaps, features, or system centralization
• design-spec-extraction — Extract production-ready JSON design specs from visual inputs using a 7-pass architecture
• devops-engineer — Dockerfiles, CI/CD pipelines, Kubernetes manifests, and Terraform/Pulumi templates
• drizzle-orm — Drizzle ORM best practices -- schemas, queries, mutations, transactions, migrations
• drizzle-postgres — PostgreSQL and Drizzle ORM best practices for type-safe database apps
• electron-builder — Electron packaging, code signing, auto-updates, and release workflows
• electron-dev — Electron development with Electron Vite and Builder -- main/renderer processes, IPC
• electron-release — Electron production builds, notarization, auto-updates, and releases
• elysia — Type-safe APIs with Elysia including routing, validation, plugins, and error handling
• es-toolkit — Modern utility library as a lodash replacement -- array, object, string operations
• evolution-api — Evolution API for WhatsApp messaging, instance management, and chatbot orchestration
• executing-plans — Execute implementation plans in batches with review checkpoints
• find-rules — Discover project rules, coding standards, and architectural guidelines
• find-skills — Discover and install agent skills from the open agent skills ecosystem
• fixing-motion-performance — Diagnose and fix animation performance issues in UI code
• helm-chart-scaffolding — Design, organize, and manage Helm charts for Kubernetes applications
• hetzner-server — Create and manage Hetzner Cloud servers via the hcloud CLI
• inngest — Serverless background jobs, event-driven workflows, and durable execution
• k8s-security-policies — Kubernetes security policies -- NetworkPolicy, PodSecurityPolicy, and RBAC
• kubernetes-specialist — Deploy and manage Kubernetes workloads -- manifests, Helm, RBAC, networking, GitOps
• mermaid-diagrams — Software diagrams using Mermaid syntax -- class, sequence, flowcharts, ERD, C4
• motion — React animations with Motion (formerly Framer Motion) -- gestures, scroll effects, spring physics
• motion-react — Full Motion for React guide including SVG, exit animations, and layout transitions
• obsidian-bases — Create and edit Obsidian Bases (.base files) with views, filters, formulas, and summaries
• obsidian-cli — Interact with Obsidian vaults via CLI -- read, create, search, manage notes, and develop plugins
• obsidian-markdown — Obsidian Flavored Markdown with wikilinks, embeds, callouts, properties, and tags
• pal — Pal MCP toolkit for code analysis, debugging, planning, refactoring, and tracing
• perplexity — Web search and research using Perplexity AI
• remotion-best-practices — Best practices for Remotion video creation in React
• requirements-clarity — Clarify ambiguous requirements through focused dialogue before implementation
• ship-learn-next — Transform learning content into actionable implementation plans
• sourcebot — Search external libraries and frameworks using Sourcebot MCP
• tauri-v2 — Tauri v2 cross-platform apps with Rust backend, IPC, permissions, and builds
• terraform-style-guide — Terraform HCL following HashiCorp's official style conventions
• workflow — Durable, resumable workflows using Vercel's Workflow DevKit
• wrangler — Cloudflare Workers CLI for deploying and managing Workers, KV, R2, D1, and more

Structure

Each skill follows a consistent directory layout:

skills/
  mine/<skill-name>/       # Original skills authored here (⭐️)
  curated/<skill-name>/    # Hand-picked community skills (💎)
  marketing/<skill-name>/  # Marketing, business, and writing skills (📣)
  community/<skill-name>/  # Broader community skills

skills/<bucket>/<skill-name>/
  SKILL.md              # Main skill definition (required)
  references/           # Deep-dive reference material
  examples/             # Usage examples and patterns
  templates/            # Code templates and scaffolds
  scripts/              # Automation scripts and validators
  checklists/           # Step-by-step verification checklists

Contributing

To add a new skill:

1. Create a directory under the appropriate bucket with a lowercase, hyphenated name:
   • skills/mine/ — original work authored here
   • skills/curated/ — hand-picked, high-quality community skills you maintain
   • skills/marketing/ — marketing, sales, business, or writing skills
   • skills/community/ — broader community skills
2. Add a SKILL.md with proper frontmatter (name and description fields)
3. Include reference material, examples, and templates as needed
4. Follow the conventions documented in skills/mine/writing-skills/SKILL.md

License

See repository root for license information.