criar-skill

criar-skill is a skill for Claude Code from pe-menezes/pm-chief-of-staff. It costs 125 tokens per session (1,461 once invoked), scanned A, original, MIT.

A process for turning a repeatable personal workflow into a custom skill for a coding agent.

In plain words
What is it for?
Use it to automate a recurring process, test whether the agent recognises its trigger, and plant the finished skill in the person's project.
Why use it?
It first checks whether the request is really a reusable workflow rather than a general preference or one-off rule, then guides writing, testing, and installing the skill.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions Claude Code.

Part of the pm-chief-of-staff plugin — 10 skills, 1 hook shipped together

Good fit Use it to automate a recurring process, test whether the agent recognises its trigger, and plant the finished skill in the person's project.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/pe-menezes/pm-chief-of-staff/criar-skill
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add pe-menezes/pm-chief-of-staff --skill criar-skill
Clone the repo
git clone --depth 1 https://github.com/pe-menezes/pm-chief-of-staff

Made for: Claude Code.

Or install pm-chief-of-staff, the plugin that ships this one along with the rest of its 10 skills, 1 hook.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for criar-skill

README.md
[![agentmods](https://agentmods.dev/badge/skills/pe-menezes/pm-chief-of-staff/criar-skill/github.svg)](https://agentmods.dev/skills/pe-menezes/pm-chief-of-staff/criar-skill)
Your own site
<a href="https://agentmods.dev/skills/pe-menezes/pm-chief-of-staff/criar-skill"><img src="https://agentmods.dev/badge/skills/pe-menezes/pm-chief-of-staff/criar-skill/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for criar-skill

Your own site · 80×15
<a href="https://agentmods.dev/skills/pe-menezes/pm-chief-of-staff/criar-skill"><img src="https://agentmods.dev/badge/skills/pe-menezes/pm-chief-of-staff/criar-skill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 125 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,461 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00125 $0.01461
Opus 5 $0.00063 $0.00731
Sonnet 5 $0.00025 $0.00292
Haiku 4.5 $0.00013 $0.00146

Measured 12d ago against content hash 3b104c883631, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

criar-skill scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/criar-skill/SKILL.md · 47 lines

How it starts

The opening of the file, as written. The whole thing — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Criar uma skill

A pessoa pediu pra você automatizar um processo. O caminho tem 4 passos: triar se aquilo vira skill mesmo, escrever a skill, testar o disparo com ela e plantar na pasta dela. Leia esta skill inteira antes de propor qualquer coisa.

Passo 1: triagem (nem todo pedido vira skill)

Vira skill só o workflow repetível com gatilho próprio: um processo com passos e resultado, que a pessoa vai invocar de novo e que se reconhece numa frase ("processa a retro", "monta o report do comitê"). Pedido que não passa nesse filtro tem outro destino:

  • Preferência que vale em toda interação (tom, formato de resposta, vocabulário): 1 linha no CLAUDE.md.
  • Regra situacional de um tema ou tipo de arquivo ("quando mexer em nota de reunião, faça X"): arquivo de regra que só carrega quando o trabalho toca aqueles arquivos, e o CLAUDE.md continua leve (a mecânica está na camada de implementação, adiante).
  • Disciplina de um passo de execução que 2 ou mais skills precisam ("antes de escrever a nota, aplique isto"): módulo compartilhado em .claude/skills/_shared/{nome}.md (mecânica na camada de implementação, adiante). Se só uma skill precisa, é arquivo na pasta daquela skill. E .claude/rules/ não substitui o módulo: regra com paths: dispara quando um arquivo é tocado, e disciplina de "antes de escrever" tem que carregar no meio da execução, antes de o arquivo existir.
  • Ajuste em skill que já existe (semente plantada ou skill criada antes): edição direta daquela skill. Skill nova com gatilho vizinho de uma existente disputa o disparo com ela.

Diga qual destino o pedido tem e por quê, e siga pro passo 2 só com a confirmação da pessoa.

Passo 2: escrever a skill

Skill é instrução pra agente, e as restrições que valem pra qualquer instrução valem aqui: contexto é finito, instrução longa degrada qualquer modelo, gatilho vago não dispara em agente nenhum. As regras de autoria saem dessas restrições:

  • description em 3ª pessoa: o QUE + QUANDO. O que a skill faz, quando usar (com os termos que a própria pessoa usaria pra pedir) e quando NÃO usar, apontando o destino certo. A description decide o disparo e o corpo só é lido depois: gatilho vago não dispara.
  • Corpo é guia leve que confia no julgamento do modelo. Objetivo, etapas e as decisões de formato da pessoa, sem prescrever o que o agente resolve sozinho: instrução longa degrada a execução inteira. Regra rígida só em passo frágil ou destrutivo (sobrescrever, apagar, mandar algo pra fora), e sempre com o porquê: regra sem porquê quebra no primeiro caso que ela não previu, e o porquê é o que orienta o caso novo.
  • Progressive disclosure. Corpo curto; template longo, tabela de referência ou detalhe que só serve em parte das execuções vai pra arquivo separado na mesma pasta da skill, que o corpo aponta. O detalhe só deve ocupar contexto quando a execução precisa dele.
  • Artefato tem alvo em número. Skill que gera arquivo declara a mira ("nota de até 25 linhas", "1 linha por item"); sem número o output cresce a cada execução.
  • Sem exemplos de uso engessados. Diálogo pronto e output modelo constrangem a exploração: o agente passa a imitar o exemplo em vez de resolver o caso real.
  • A skill codifica a opinião e o jeito DA PESSOA. O formato de pauta dela, o vocabulário dela, o que ela considera pronto. Processo genérico que serviria pra qualquer um não carrega opinião e não precisa de skill.

Read the full file on GitHub · 47 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 12d ago First seen · 47 lines · 0 tokens per session scan A 3b104c883631

Subscribe to this mod's changes

criar-skill is a skill published in the GitHub repository pe-menezes/pm-chief-of-staff (2 stars, last pushed 1mo ago), licensed MIT. It adds 125 tokens to every session and 1,461 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

reduce

Extract structured knowledge from source material. Comprehensive extraction is the default — every insight that serves the domain gets extracted. For domain-relevant sources, skip rate must be below 10%. Zero extraction from a domain-relevant source is a BUG. Triggers on "/reduce", "/reduce [file]", "extract…

agenticnotetaking/arscontexta · 73 tokens

migrate

Scan the project and migrate existing documentation into OKF format. Use when the user wants to populate the knowledge base from existing docs, README, schemas, API specs, runbooks, or any structured project knowledge. Trigger on "migrate to okf", "populate the knowledge base", "scan my docs", "convert docs to okf"…

guhcostan/claude-mega-brain · 147 tokens

rethink

Challenge system assumptions against accumulated evidence. Triages observations and tensions, detects patterns, generates proposals. The scientific method applied to knowledge systems. Triggers on "/rethink", "review observations", "challenge assumptions", "what have I learned".

agenticnotetaking/arscontexta · 51 tokens

ralph

Queue processing with fresh context per phase. Processes N tasks from the queue, spawning isolated subagents to prevent context contamination. Supports serial, parallel, batch filter, and dry run modes. Triggers on "/ralph", "/ralph N", "process queue", "run pipeline tasks".

agenticnotetaking/arscontexta · 61 tokens

remember

Capture friction as methodology notes. Three modes — explicit description, contextual (review recent corrections), session mining (scan transcripts for patterns). Triggers on "/remember", "/remember [description]".

agenticnotetaking/arscontexta · 40 tokens

verify

Combined verification — recite (description quality via cold-read prediction) + validate (schema compliance) + review (health checks). Use as a quality gate after creating notes or as periodic maintenance. Triggers on "/verify", "/verify [note]", "verify note quality", "check note health".

agenticnotetaking/arscontexta · 61 tokens