Pedro Nauck's Skills

A curated collection of 87 agent skills for Claude Code and compatible AI coding assistants — 40 original (⭐️), 34 hand-picked (💎), and 13 marketing & business (📣). 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 three top-level buckets:

• skills/mine/ — 40 original skills authored in this repository (⭐️)
• skills/curated/ — 34 hand-picked skills, including locally authored skills (💎)
• skills/marketing/ — 13 marketing, business, and writing 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
• ddd-master — Assess DDD fit and model language, bounded contexts, and aggregates using scoped references and existing domain evidence
• deep-review — Review branch diffs, working trees, and PRs with automated preparation, staged rule discovery, compact per-job assessments, targeted repairs, complete defect/polish coverage, and optional PR publication
• 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
• glassmorphism — Glass and Apple Liquid Glass interfaces on the web at reference quality: a material anatomy (thin light fill, one locked light direction expressed as rim + inset highlight + sheen + tinted shadow, optional grain), three ink modes chosen by backdrop luminance (white on light tint, white on smoked tint, dark ink on white frost) with measured contrast, a solid-first glass.css token system with reduced-transparency / forced-colors / print fallbacks, geometry-derived SVG displacement refraction for small controls (Chromium-only, with the filter-region and chromatic-pass traps), a runnable material lab with seven backdrops and a live contrast estimate, a 22-item reference board, rendering-cost budgets and triage, and conditional SwiftUI / visionOS / Fluent / Expo routes
• 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 scoped dogfooding through public product interfaces, record findings in living QA docs, and repair within authorization; assess PR or release readiness when included in the request
• 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 — Choose owning test suites and independent behavioral oracles, use mocks at unit I/O boundaries, validate the relevant real integration, and diagnose flaky tests or agent evals without mandatory companion-test quotas
• to-prompt — Create a handoff brief in docs/prompts/ with relevant evidence, accepted user decisions, and unresolved implementation choices
• 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 — Design and review UI using project authorities, observable usability and accessibility evidence, conditional design/performance references, and optional audit templates
• 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 — Write and audit AGENTS.md/CLAUDE.md instructions: scope project rules, preserve canonical ownership, and make workflow authorization and stopping conditions explicit
• writing-skills — Create, refactor, and debug agent skills: concise discovery descriptions, conditional reference loading, package authoring, static metadata checks, and diagnosis of missed references
• writing-tech-post — Draft and edit engineering posts with supported claims, conditional structure/voice references, and publication checks scoped to the requested artifact
• 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 skills maintained in this repository, including locally authored skills placed here by the owner.

• agent-browser — Automate browser interactions for testing, form filling, and data extraction
• ai-sdk — Vercel AI SDK for building AI-powered features
• autoresearch — Autonomously optimize any skill by running evals, mutating prompts, and keeping improvements
• centrifugo — Centrifugo real-time messaging -- WebSocket PUB/SUB, channels, JWT auth, scaling
• 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
• 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
• es-toolkit — Modern utility library as a lodash replacement -- array, object, string operations
• extreme-software-optimization — Profile-driven performance optimization with behavior proofs, opportunity scoring, and isomorphism guarantees
• hono — Hono framework development with documentation search and API reference
• inngest — Serverless background jobs, event-driven workflows, and durable execution
• lesson-learned — Extract software engineering lessons from git history and recent code changes
• mermaid-diagrams — Software diagrams using Mermaid syntax -- class, sequence, flowcharts, ERD, C4
• next-best-practices — Next.js best practices -- file conventions, RSC boundaries, data patterns, async APIs, metadata, error handling, and optimization
• 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
• qmd — Search markdown knowledge bases, notes, and documentation using QMD
• requirements-clarity — Clarify ambiguous requirements through focused dialogue before implementation
• 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
• tauri-v2 — Tauri v2 cross-platform apps with Rust backend, IPC, permissions, and builds
• 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
• 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
• writing-clearly-and-concisely — Strunk's timeless rules for clearer, stronger, more professional prose

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 (📣)

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

Updating imported skills

Third-party skills are tracked in upstream/sources.json, including original repositories, renamed skills, and sources that are retired or still unverified. Keep their published files in skills/ and use the maintainer command:

make check
make update
# Or update one skill:
make update SKILL=context7

Use make status for the offline inventory, make check BUCKET=curated to select a bucket, and make to see all shortcuts.

This runs the native npx skills add in isolated staging projects, preserves bucket paths and published names, and protects subsequent local edits. It uses separate catalogue locks under upstream/; the root development-environment lock is untouched. See the maintenance guide for setup, backups, source exceptions, and adding a new tracked skill. Bare npx skills check is not a read-only catalogue check in the pinned CLI version.

Adding skills

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
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.