12 skills oficiais de IA e ferramentas confiáveis | AI Agent Skills
Compare skills oficiais de IA e ferramentas confiáveis e siga direto para a documentação de instalação, a validação por CLI e os caminhos de rollout para Claude Code, Cursor e Windsurf.
Esta coleção reúne skills oficiais de IA e ferramentas confiáveis mantidas por equipes first-party e por players sólidos do ecossistema. Ela é voltada para times que precisam de qualidade previsível, documentação clara e caminhos de integração prontos para produção. MCP pode atuar como infraestrutura de compatibilidade, mas a prioridade está em skills e ferramentas práticas para acelerar entregas de IA com confiança.
Page-Level Review Standard
This Page Is a Curated Decision Surface, Not Just a Themed List
A collection page should not just add more cards to the screen. It should explain why these skills belong together, how the next step moves into installation and validation, and which path should continue the decision afterwards.
Reviewed On
2026-04-16
Maintained By
Maintained through Killer-Skills editorial review with trust, install-path, and operator checks.
Verification
Validate ownership, doc freshness, install instructions, and workflow relevance before keeping or adding an entry.
Primary Audience
High-intent users who roughly know the direction and need a faster path to an installable shortlist.
Three Minimum Standards For This Collection
- A collection page must narrow users into a better shortlist instead of flattening more repositories onto the screen.
- The next click should continue into installation docs, CLI validation, or a better-fit solution page instead of resetting back to a broad directory.
- This page only becomes a real first-party judgment surface when its selection logic, maintenance posture, and delivery path are all visible.
Primary Install Bridge
Pick One Skill, Then Take the Install Path
This collection should not trap users in comparison mode or pretend to install the whole collection. Its job is to narrow the shortlist to one skill, then send the next click into installation, validation, and rollout. Installation happens on the skill path.
Start With The Workflow Bottleneck You Actually Need To Solve
A high-intent collection should not dump every tool into one flat wall. Decide whether your bottleneck is execution scaffolding, handoff governance, or repeatable rollout first, then enter the track that matches the job you need to solve.
Track 1
Start with the safest first-party anchors
Choose this lane when the main question is trust, ownership, and install clarity. It is the fastest path for teams that want first-party defaults before they expand into broader ecosystem tooling.
Use it when a shared team stack still needs a low-risk default and you do not want the first install to depend on guesswork.
Confirm Before You Commit
- Ownership is obvious from the maintainer or product team.
- Install docs and release posture are visible before rollout starts.
- The tool can become a safe default without another wide repo crawl.
Choose this lane when the tool itself matters less than the operational guardrails around it. The goal is to lock review, context, and execution discipline before you standardize across a team.
Use it when installs keep working technically, but quality, ownership, or operator confidence still drift between projects.
Confirm Before You Commit
- The team can audit what context, review rules, and security checks surround the install.
- Operators know which tool governs review, orchestration, and failure boundaries.
- The stack is clear enough to survive more than one contributor or repo.
Choose this lane after trust is clear and the first install already works. It helps you widen the stack into framework, platform, and workflow builders without losing the original trust baseline.
Use it when you are not choosing the first safe install anymore, but deciding which trusted builders deserve a place in the next workflow lane.
Confirm Before You Commit
- The first trusted install already succeeded and is documented.
- The next tool expands capability instead of resetting the stack choice from scratch.
- You can still map the tool into a concrete workflow after installation.
The Next Click Should Keep Narrowing, Not Reset Back To A Generic Directory
Once the install path is clear, move into the solution, CLI, or editorial surface that best matches this collection. That keeps platform, framework, and operations demand narrowing into a more verifiable high-intent journey.
Reviewed on 2026-04-16 for first-party ownership, documentation quality, install clarity, and production relevance. This is the safest collection to use as a default starting point.
We prioritize this page because it lets users verify trust first and then move into one clear installation path instead of bouncing across more repo lists.
Trust Signals
- Entries are maintained by first-party product teams or widely trusted ecosystem builders.
- Documentation and install paths are public, active, and easy to validate before rollout.
- Selection favors practical deployment value over generic GitHub popularity.
Grouping Logic
- Lead with first-party tools and ecosystem anchors users can verify without guesswork.
- Keep the shortlist compact enough to compare quickly, but broad enough to cover install, orchestration, observability, and workflow automation.
- Route discovery into one installation path before expanding into broader workflow collections.
O dispatching de agentes paralelos permite resolver problemas independentes de forma eficiente, melhorando a produtividade e reduzindo o tempo de resolução.
Transformers.js permite executar modelos de aprendizado de máquina de ponta em JavaScript, tanto em navegadores quanto em ambientes Node.js, sem necessidade de servidor.
Stitch Loop is an AI agent skill that enables autonomous frontend builders to iteratively develop websites using Stitch. It integrates pages into site structures and prepares instructions for the next iteration, streamlining AI coding workflows for developers.
Guide for implementing oRPC contract-first API patterns in Dify frontend. Trigger when creating or updating contracts in web/contract, wiring router composition, integrating TanStack Query with typed contracts, migrating legacy service calls to oRPC, or deciding whether to call queryOptions directly vs extracting a helper or use-* hook in web/service.
The security-review AI agent skill detects vulnerabilities in code, providing systematic review and confidence-based reporting. It helps developers identify security issues and protect their applications.
Helps users discover and install agent skills when they ask questions like how do I do X, find a skill for X, is there a skill that can..., or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
If This Page Is Close, Keep Narrowing With Related Authority Collections
Do not reset back to the generic directory. Move sideways through these adjacent high-intent collections to narrow the shortlist toward the install path that best matches your team.
This collection should not keep users browsing forever. These three questions explain how to shortlist, install, and validate the next step.
Para quais fluxos de trabalho essas coleções foram criadas?
Essas coleções são construídas em torno de automação de fluxos, processos, documentos, dados e pilhas de habilidades reutilizáveis.
Qual a diferença entre uma coleção e o diretório principal?
O diretório é melhor para busca direta, enquanto coleções ajudam a encontrar habilidades complementares agrupadas por fluxo completo.
Posso instalar essas coleções para Claude Code ou Cursor?
Sim. As habilidades dessas coleções geralmente funcionam no Claude Code, Cursor, Windsurf e outros ambientes com instalação unificada.
Additional Next Paths
Use These Additional Paths If You Need One More Step To Narrow The Decision
These are the supporting surfaces for this collection after the install direction is clear and the primary next paths have already narrowed the decision.
