convez.ai
MCP ServerAutomaçãoIntermediário
runtypelabs

Runtype Skills

runtypelabs/skills

Skills oficiais para agentes IA no ecossistema Runtype

Ver no GitHub
BomPluginScore · de 0 a 100

Coleção oficial de skills para agentes IA na plataforma Runtype, permitindo integração com Claude Code, Cursor, Copilot e 35+ outros agentes de codificação. Inclui skills para construir produtos, administrar contas, gerenciar personas, templates e workflows.

Instalação

$npx convez-ai runtypelabs/skills

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
75%
Último commit
há 6 dias
Versão
—
Issues abertas
17
Contribuidores
2

Casos de uso

  • Construir e validar produtos Runtype com agentes e fluxos de trabalho
  • Administrar contas Runtype com segurança através de MCP
  • Embutir e tematizar widgets de chat Persona em aplicações
  • Criar e distribuir templates Runtype reutilizáveis
  • Integrar SDK Runtype, CLI e workflows em agentes de IA

Veredito

Skills bem estruturados e documentados da Runtype para automatizar workflows, administração e construção de produtos. Abordagem profissional com validação, lint e práticas recomendadas, mas requer familiaridade com o ecossistema Runtype para máxima utilidade.

README

A tradução para o português deste README ainda está na fila.

Runtype Skills

skills.sh

Official agent skills for the Runtype platform — install them in Claude Code, Cursor, Copilot, and 35+ other AI coding agents.

Install

All skills:

npx skills add runtypelabs/skills

Pick specific skills:

npx skills add runtypelabs/skills --list

Skills

SkillDescription
runtypeUmbrella onboarding and routing skill for Runtype setup, capability scoping, and choosing the right focused workflow.
runtype-build-productBuild and validate hosted Runtype products with agents, flows, tools, surfaces, records, schedules, secrets, and evals.
runtype-adminOperate live Runtype accounts safely through MCP or Code Mode MCP: inspect resources, debug traces, logs, evals, and apply careful mutations.
runtype-personaEmbed, theme, and debug Persona chat widgets, fullscreen assistants, client tokens, artifacts, and browser-side local tools.
runtype-templatesCreate, validate, and package distributable Runtype FPO templates with pending secrets and import readiness checks.
runtype-sdk-marathonUse the Runtype SDK, CLI, Marathon, playbooks, sandboxes, and code-first stored/upsert/virtual workflows.

Validation

Run the repository skill lint before publishing:

node scripts/lint-skills.mjs
node scripts/smoke-install-skills.mjs

The lint checks frontmatter, description length, linked references, forbidden internal URLs, common secret leaks, semantic guardrails for drift-prone claims, and extra drift against the neighboring ../core monorepo when it is present. The smoke test copies each skill into an isolated install directory and confirms standalone metadata and local references still resolve.

For the recommended private-source/public-publish workflow, see Keeping Runtype Skills in Sync.

Authoring a New Skill

Create a directory under skills/ with a SKILL.md file:

skills/
└── my-skill/
    ├── SKILL.md           # Required: frontmatter + instructions
    ├── scripts/           # Optional: executable helpers
    ├── references/        # Optional: API specs, schemas, docs
    └── examples/          # Optional: example inputs/outputs

SKILL.md format

---
name: my-skill
description: >-
  What it does and when to use it. Include trigger phrases.
  Max 1024 chars.
user-invocable: true
argument-hint: "[optional args hint]"
---

Instructions for the agent go here. Keep under 500 lines.
Move detailed reference material to references/ or scripts/.

Best practices

  • Description: Use imperative phrasing ("Use when..."), include specific trigger phrases
  • Size: Keep SKILL.md under 500 lines / 5,000 tokens
  • Structure: Move API specs, schemas, and long docs to references/
  • Scripts: Bundle reusable helpers in scripts/ instead of inlining
  • No secrets: Never include API keys, internal URLs, or credentials

License

MIT

Ver o original no GitHub

Bom para

Relacionados