Explore AI Agent Skills & Claude Prompts
Discover open-source agent skills for Claude Code, Codex, ChatGPT, and any tool that uses SKILL.md.
Enter through keywords, occupations, creators, and GitHub sources to see what kinds of skills are emerging across domains.
Use the same catalog through the API
Connect 381,784 public skills to your own search, analytics, or agent workflow with the REST API.
Querying local SQLite index...
design-deep-analysis
by wagnerra23ATIVAR quando Wagner pedir /design-deep <persona-slug>, "analisar visualmente tela X pra persona Y", "design profundo da tela <Z>", OU em refator visual de tela que afeta cliente paga + reportou fricção. Skill canônica de análise contextualizada por persona — carrega persona YAML de memory/clientes/<cliente>/personas/, invoca design:* skills Anthropic em paralelo (critique + system + ux-copy + accessibility-review + research-synthesis), score 15 dimensões com ponderação por persona, entrega 3 alternativas A/B/C com diff de código preparado + métrica antes/depois. Refs ADR UI-0016, framework-15-dimensoes.md, RUNBOOK-design-deep.md.
mwart-comparative
by wagnerra23Use SEMPRE antes de codar Page Inertia em migração MWART (Blade→React) no oimpresso. Skill Tier A always-on V4 que ORQUESTRA o Claude Design plugin Anthropic (design:design-critique + design:design-handoff + design:design-system + design:ux-copy + design:accessibility-review + design:research-synthesis) **e** o loop Cowork ↔ Claude Code formalizado em `prototipo-ui/` (ADR 0114). Gera artefato OBRIGATÓRIO `memory/requisitos/<Mod>/<tela>-visual-comparison.md` com 15 dimensões + framework Anthropic completo + sincroniza com `prototipo-ui/SYNC_LOG.md`. Skill PARA após gerar draft e aguarda Wagner aprovar SCREENSHOT (não tabela) ~10min síncrono ANTES de qualquer Edit/Write em `resources/js/Pages/<Mod>/<Tela>.tsx`. Restaura o loop "design supervisionado" da era Repair S2.5 com qualidade estado-da-arte. Ativa quando user pede "migrar tela X pra MWART", "comparativo visual", "/mwart-comparative <tela>", OU em qualquer Edit/Write em Page Inertia que não tenha visual-comparison.md ao lado.
mwart-process
by wagnerra23Use SEMPRE que o trabalho envolva migrar tela Blade legacy → Inertia/React no oimpresso (MWART). Carrega o processo canônico ÚNICO definido em ADR 0104 — 5 fases obrigatórias e sequenciais (PLAN → BACKEND BASELINE → FRONTEND INCREMENTAL → QA → CUTOVER). Não há caminho alternativo. Ativa quando o pedido é "migrar tela X pra MWART", "criar tela em Pages/<Mod>/<Tela>.tsx", "migrar Blade pra React", ou quando Edit/Write em qualquer `resources/js/Pages/<Mod>/<Tela>.tsx` ou em controller chamando `Inertia::render`.
mwart-quality
by wagnerra23Use ANTES de criar/editar tela MWART (Module Web App React Transition Blade→Inertia/React) no oimpresso. Ativa quando user pede "migrar tela X pra MWART", "S2.5/S2.6 tela", "nova tela Inertia em Modules/", OU em qualquer Edit em `Modules/<X>/Http/Controllers/*Controller.php` que chama `Inertia::render(...)`, OU em `resources/js/Pages/<Module>/**/*.tsx`. Carrega 9 pré-flight checks que evitam os 5 padrões de bug recorrentes detectados nos PRs
criar-staging
by wagnerra23ATIVAR quando user pedir "criar staging", "ambiente de homologação/homolog", "replicar produção pra teste", "subir/recriar/atualizar staging.oimpresso.com", "clone da produção pra equipe testar", "ambiente de teste antes de mexer em produção", "re-seedar o staging", OU qualquer Edit/Write em `docker/oimpresso-staging/**`. Carrega o processo canônico de criar/manter um clone FIEL da produção (ERP web inteiro) no CT 100 — subdomínio próprio + banco MariaDB dedicado ANONIMIZADO (LGPD) + integrações NEUTRALIZADAS — sem precisar perguntar pra ninguém como faz. Aponta pro RUNBOOK detalhado + scripts versionados + as 10 pegadinhas já catalogadas. Origem 2026-05-29 (Wagner: "criar uma regra pro mcp saber fazer isso sem eu precisar informar ninguém").
design-memoria-reprocess
by wagnerra23ATIVAR quando (a) o Claude Design enviar handoff com bloco `## new_design_memories`; (b) um doc de design for criado/editado/aposentado (`prototipo-ui/*.md`, `memory/requisitos/_DesignSystem/*.md`, `*.charter.md`); (c) um ADR UI novo com `supersedes` for mergeado; (d) novo golden de arquétipo / bump de token (DS v5) / mudança de método; OU (e) user pedir "reprocessar índice de design", "atualizar memórias de design", "destravar fluxo de design", "/design-memoria-reprocess". Mantém o INDEX-DESIGN-MEMORIAS.md vivo SEM quebrar a estrutura — append-only + ratchet + freshness + idempotente. Tier B auto-trigger. Refs ADR proposto governanca-evolucao-doc-design, ADR 0230/0231/0233, INDEX-DESIGN-MEMORIAS.md.
charter-first
by wagnerra23BLOQUEADOR — ANTES de editar qualquer .tsx que tenha .charter.md ao lado (ex Index.tsx + Index.charter.md), chame tool MCP `charter-fetch <page-id>` pra carregar contrato vivo da página (Mission/Goals/Non-Goals/UX targets/Anti-hooks). Tier A always-on — princípio duro
charter-write
by wagnerra23ATIVAR quando user pedir "criar charter da tela X", "escrever charter pra /caminho", "gerar charter de Index.tsx Y", "novo charter Page", "/charter-write {pagina}". Lê o `.tsx` da tela + Controller + topnav/routes pra inferir Mission/Goals/UX targets/Automation hooks; gera draft em `*.charter.md` ao lado do `.tsx`; PARA aguardando Wagner revisar Non-Goals + Anti-hooks (parte mais sensível, anti-alucinação). NUNCA marca charter como `status: live` sozinho — Wagner aprova.
incident-done-checklist
by wagnerra23BLOQUEADOR — ATIVAR antes de declarar "incident fechado" / "está pronto" / "feature funcionando" / encerrar sessão de fix em prod. Skill carrega a Definition of Done canônica (DoD-v1) que EXIGE smoke real prod end-to-end pra cada fix antes de marcar pronto. Funciona como gate procedural — sem TODOS os checks ✅, status fica `awaiting-smoke` no commit/PR/handoff, NÃO `done`. Aprende incident 2026-05-28 onde declarei "10 PRs fechados" com 3 fixes (M1/M2/M3 mídia) NUNCA validados por smoke real → 10.144 mídias ainda órfãs prod descoberto pelo Wagner depois. Operacional Tier A — DEVE ativar SEMPRE que agente escreve "está pronto", "fechado", "completou", "deployed", "validado" em mensagem ao Wagner. Bloco D (Reflexion runtime): quando o incidente foi erro de OPERAÇÃO da Jana (≠ saída do LLM), registrar a lição em Modules/Jana/LICOES-OPERACAO.md + graduar (MEC→check jana:health-check / JULG→regra). Refs PATTERN-INCIDENT-RESPONSE-VELOCITY.md passo 4, ADR 0093, skill commit-discipline.
cowork-prototype-replication
by wagnerra23ATIVAR quando user pedir "fazer layout estado-da-arte", "replicar protótipo Cowork", "espelhar visual-source.html", "transformar prototipo-ui/* em Inertia React", "usar layout do cockpit pra módulo X", OU em Edit/Write em `resources/js/Pages/<Mod>/<Tela>.tsx` quando existe `prototipo-ui/prototipos/<tela>/visual-source.html` ou `F1.html` correspondente. Carrega processo canônico de 7 fases (F0 sync + F1 mapping vocabulário vertical + F2 mapping CSS Cowork→Tailwind + F3 component hierarchy + F4 useMemo/useCallback + F5 Pest + F6 deploy + F7 smoke INTERATIVO) — RUNBOOK detalhado em [memory/requisitos/_DesignSystem/RUNBOOK-replicar-prototipo-cowork.md](../../memory/requisitos/_DesignSystem/RUNBOOK-replicar-prototipo-cowork.md). Caso real validado: Kanban Producao Oficina Caçambas 2026-05-13 (PRs #735→#740 madrugada pré-Martinho 10h).
wagner-request-refiner
by wagnerra23ATIVAR quando Wagner manda múltiplos pedidos curtos não-estruturados num mesmo turno (ex: lista com 3+ items, "todo: a) b) c)", bullets numerados, screenshots com várias anotações simultâneas, ou texto corrido com várias intenções misturadas). Decompõe em tasks atômicas, infere owner/priority/module/estimate cruzando com SPEC.md + MCP, propõe estrutura ANTES de criar via MCP ou editar código. Anti-pattern: pegar tudo de uma vez e implementar sem cruzar dependências. Wagner valoriza economia de crédito + escopo confirmado antes de execução massiva.
copiloto-arch
by wagnerra23Use ao trabalhar em Modules/Copiloto/ ou ao tocar memória/IA do projeto. Carrega arquitetura canônica do Copiloto (ADRs 0035-0053): laravel/ai SDK, MeilisearchDriver hybrid, ContextoNegocio com 3 ângulos faturamento, OTel GenAI, MCP server governança, tabela copiloto_memoria_metricas. Substitui leitura repetida de 18+ ADRs.
Browse Agent Skills by Occupation
23 major groups · 867 SOC occupations
Browse by Category
Explore agent skills organized by their primary use case
Explore the agent skills ecosystem by occupation and creator
SkillMD is not just a keyword search box. It is an open map that organizes public skills by occupation, creator, and repository, helping you see which workflows, judgment criteria, and domain habits people are writing for AI agents.
Then follow creators and GitHub repositories back to the source: compare the skills a team maintains, whether the repo is active, and how the README frames the work before you open, install, or reuse anything.
Use it three ways: learn an unfamiliar field by occupation, study how creators organize skills, then use source context to decide what is worth opening or reusing.
01 Map a field
Browse 23 occupation groups and 867 SOC roles to learn what skills exist in adjacent domains and how they break down real work.
02 Follow creators
Use creator and repository pages to inspect maintained skill collections, recent updates, and source context before trusting a result.
03 Search with sources
Search 1.7M+ collected skills, then use occupation tags, creators, and GitHub source context to decide what is worth opening.
Start with the occupation map, then follow creators and repositories back to real code. SkillMD helps explain why a skill is worth opening, not only what it is named.
Standardizing Agent Capabilities with SKILL.md and Model Context Protocol (MCP)
In the rapidly evolving landscape of artificial intelligence, LLM agents (Large Language Model agents) have transitioned from simple text predictors to autonomous problem solvers. To orchestrate complex, multi-step agentic workflows, developers require a standardized format to specify agent capabilities, prompt instructions, system rules, and database bindings. This is where SKILL.md and the Model Context Protocol (MCP) have emerged as standard developer paradigms. SkillMD serves as the central directory for indexing, exploring, and sharing these critical agent configurations.
Our open-source registry currently tracks over 1.7 million collected SKILL.md configurations and system prompts. By compiling agent configurations from active developers on GitHub, we bridge the gap between prompt engineering research and production execution. Whether you are building agents with Anthropic's Claude Code, OpenAI's GPT-4, Google's Gemini, or local models using Ollama and LlamaIndex, standardized skill definitions ensure your agents behave predictably across different runtime environments.
What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open-source standard designed to connect LLMs to data sources, developer tools, and external environments. MCP establishes a bidirectional communication channel between client applications (like Cursor, Claude Desktop, or custom agent systems) and servers hosting data or capabilities. Standardizing instructions via SKILL.md enables LLMs to query databases, read local files, execute terminal commands, and integrate third-party APIs. SkillMD allows you to find ready-to-run MCP servers and prompt instructions for various occupations and technical tasks.
The Structure of a Professional SKILL.md File
A valid SKILL.md configuration is designed to be easily read by humans and parsed by LLMs. It contains precise system instructions, trigger conditions, required parameters, and execution examples. Below is the typical architectural blueprint of a professional agent skill:
- Metadata & Core Scope: Declares the name of the skill, author details, target models, and a description of the capability.
- Triggers & Intent Detection: Details semantic triggers that help the agent decide when to invoke this skill.
- System Prompts: Explicit system-level instructions that direct the agent's behavior, personality, safety guardrails, and formatting preferences.
- Capabilities & Tools: Lists the files, databases, or APIs the agent must access to complete the tasks.
- Few-Shot Examples: Demonstrates real inputs and outputs, helping the model generalize behavior through in-context learning.
Optimizing Agent Workflows for Modern LLMs
Writing effective agent skills requires deep knowledge of prompt engineering. With the release of advanced reasoning models like Claude 3.5 Sonnet, ChatGPT o1, and DeepSeek-V3, prompt templates must focus on structured thinking. Developers are encouraged to use XML tags (e.g., <thought>, <context>, and <rules>) to isolate execution boundaries. Standardized prompts prevent agents from suffering from context drift, ensuring that long-running tasks remain aligned with the initial system parameters.
Exploring by SOC Occupations and Creator Profiles
What makes SkillMD unique is its taxonomy. Instead of simple text search, we parse and organize files according to the Standard Occupational Classification (SOC) system. This means you can discover skills written for Computer and Mathematical roles, Business and Financial operations, Legal, Design, and and Educational Instruction fields. By tracking creator profiles, developers can study how different teams organize their custom instructions, compare version updates, and fork public configs for specialized enterprise use cases.
SkillMD operates as a high-performance index running on a fast Go backend and a highly responsive Astro SSR frontend. All search queries execute in milliseconds, featuring smart debouncing to prevent multiple API requests while keeping user data secure. Join our community of developers to standardize your AI agent instructions and optimize your LLM prompting workflows today.
Frequently Asked Questions
A practical guide to agent skills: what they are, how to inspect them, and how SkillMD helps you explore the ecosystem.