convez.ai
SkillAutomaçãoIntermediário
fumb13s

Around: Skills Customizados para Claude Code

fumb13s/around

Skills para desenvolvimento autônomo e monitoramento de uso em Claude Code

Ver no GitHub
InicialPluginScore · de 0 a 100

Coleção de skills Shell para Claude Code que automatiza fluxos de desenvolvimento completo (do issue ao PR) e monitora uso de tokens/custos. Inclui lightbulb para desenvolvimento autônomo end-to-end e usage para análise de consumo de API.

Instalação

$npx convez-ai fumb13s/around

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
0
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 1 mês
Versão
—
Issues abertas
3
Contribuidores
1

Casos de uso

  • Automatizar desenvolvimento completo de issues do GitHub (planejamento, implementação, review, PR)
  • Monitorar uso de tokens e custos de API em projetos Claude Code
  • Brainstorming e transformação de ideias em issues estruturados e implementados

Veredito

Skill bem construído que elimina overhead manual em ciclos de desenvolvimento. Útil para quem trabalha com Claude Code regularmente, mas requer dependências externas (superpowers) e familiarity com Git/GitHub workflow.

README

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

around

Custom Claude Code skills.

Installation

Clone the repo and run the install script:

git clone git@github.com:fumb13s/around.git ~/around
~/around/install.sh

This symlinks all skills into ~/.claude/skills/ and installs their permissions.

Options

# Install specific skills only
~/around/install.sh lightbulb
~/around/install.sh usage

# Copy instead of symlink
~/around/install.sh --copy

# Install into the current project instead of user-level
~/around/install.sh --project

# Check installation status
~/around/install.sh --check

# Remove skills and permissions
~/around/install.sh --remove

lightbulb

End-to-end autonomous development from a GitHub issue or a topic/idea. Takes an issue number -- or a plain-text idea that it brainstorms into an issue automatically -- and drives the full workflow: planning, implementation, code review, and PR creation, all via subagents in an isolated git worktree.

Flow: fetch issue -> brainstorm design -> write plan -> implement (via SDD) -> create draft PR -> review loop -> CI check -> mark ready or merge

What to expect

When you run /lightbulb 42, the skill will:

  1. Fetch the issue and create an isolated git worktree
  2. Brainstorm a design and write an implementation plan
  3. Implement the plan via subagents
  4. Open a draft PR and run review rounds
  5. Check CI and ask you how to finish (mark ready or merge)

Invocation

Issue mode

Provide a GitHub issue number to develop an existing issue end-to-end:

/lightbulb 42

Or just tell Claude: "implement issue #42" -- the skill triggers when a GitHub issue number is provided with intent for autonomous development.

Topic mode

Provide a topic or idea instead of an issue number. The skill will brainstorm a design, file it as a GitHub issue, and then develop it end-to-end:

/lightbulb add a changelog that auto-generates from PR titles

Or in natural language: "build a retry mechanism for flaky API calls" -- the skill detects that no issue number was given and enters topic mode automatically.

Options

Set max review rounds (default 5) in natural language:

implement issue #42 with at most 3 review rounds

Dependencies

Delegates to these superpowers skills:

  • superpowers:brainstorming -- design exploration
  • superpowers:writing-plans -- plan creation
  • superpowers:subagent-driven-development -- implementation

Make sure the superpowers plugin is installed in your Claude Code instance.


usage

Show Claude Code token usage, costs, and billing breakdown. Wraps ccusage with an aggregation script that produces compact daily and per-project summaries.

What it does

Fetches token usage data via ccusage, aggregates it with a bundled Python script, and presents a summary table with output tokens, cache reads/writes, totals, and API-equivalent cost.

Invocation

/usage

Or ask in natural language: "how much am I using?", "which project costs the most?", "what's my usage this week?"

Views

ViewWhen to use
Daily (default)Recent trends
WeeklyLonger-term patterns
SessionPer-conversation breakdown
BlocksBilling period / burn rate

Add per-project breakdown, single-project filter, or model-specific cost split as needed.

Dependencies

  • ccusage -- installed on demand via npx
  • Python 3 -- for the aggregation script
Ver o original no GitHub

Bom para

Relacionados