convez.ai
SkillDevOps & InfraestruturaIntermediário
erclx

AITK - Toolkit Agent-First para Desenvolvimento com IA

erclx/toolkit

Sincroniza governança e skills de IA entre projetos via CLI agent-first

Ver no GitHub
InicialPluginScore · de 0 a 100

CLI que instala e sincroniza governança, skills e padrões em projetos, com suporte para Claude Code plugins e Gemini. Funciona como fonte única de verdade para prompts, regras de agentes, comandos slash e documentação, sincronizando automaticamente entre projetos para manter consistência.

Instalação

$npx convez-ai erclx/toolkit

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
2
Forks
0
Acompanhando
0
Saúde
57%
Último commit
há 1 dia
Versão
v5.99.0
Issues abertas
4
Contribuidores
1

Casos de uso

  • Manter prompts e regras consistentes entre múltiplos projetos
  • Distribuir standards e snippets para times usando IA
  • Scaffolding de novos projetos com governança pré-configurada
  • Sincronização bidirecional de padrões entre repositórios

Veredito

Ferramenta robusta e bem-pensada para times que usam agentes de IA em múltiplos projetos. A abordagem agent-first (JSON + CLI sem prompts) e text-native (markdown legível por humanos e máquinas) resolve um problema real de drift em configurações distribuídas. Exige familiaridade com shell, Bun e fluxos Git, mas oferece automação genuína em governança de código.

README

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

aitk

Agent-first toolkit for AI-assisted development.

Every AI coding setup accumulates the same assets. Prompts to reuse, rules agents should follow, slash commands, skills, seed docs, sync scripts. Across enough projects the copies drift, and agents stop getting consistent signals.

aitk is one authoritative source for those assets and a CLI that installs and syncs them into any project. It ships a Claude Code plugin with skills for planning, review, docs sync, and the git ship chain, plus Gemini CLI commands, shared rules, snippet libraries, and seed docs.

Why

Three design choices shape the toolkit.

  • Agent-first: every command has a non-interactive path and a JSON catalog. If a Claude Code skill or any other agent cannot drive the CLI without prompts, the design is wrong.
  • Text-native: conventions, rules, and prompts are authored as markdown that humans and agents read the same way. No hidden behavior, no compiled state.
  • One source, many consumers: this repo is the authoritative copy. Target projects install and sync on demand, never author in place.

Prerequisites

  • Bun for the CLI runtime and scripts
  • Git with worktree support
  • GitHub CLI (optional) for ship flows
  • Shell: zsh or bash 4+ (brew install bash on macOS).

Quickstart

Clone, install dependencies, and link the CLI globally.

git clone https://github.com/erclx/aitk.git
cd aitk
bun install
bun link

Confirm the install.

aitk --help

Scaffold a fresh project.

mkdir ~/my-project && cd ~/my-project
git init
aitk init

aitk init installs base tooling configs, Claude seeds, standards, snippets, and a wiki stub in one pass. Pass --stack <name> to also install governance rules for a framework stack into .claude/rules/. Run aitk tooling list --json to see the catalog.

For the full outsider journey (scaffold, add a domain later, sync upstream drift), see target projects.

What is inside

Each domain has a canonical source in this repo and a thin install or sync CLI on the target side.

  • Claude Code plugin: skills for planning, review, docs sync, and the git ship chain
  • Governance rules: Cursor rules and stacks, installable per project
  • Standards: shared authoring conventions, synced to projects
  • Gemini commands: slash commands for the Gemini CLI
  • Snippets: reusable prompts for Claude and Gemini chat
  • Prompts: system prompt templates for AI authoring
  • Tooling stacks: golden configs, seeds, and references per framework
  • Design system: DESIGN.md token shape, extract skill, render command
  • Sandbox: scenario-based scaffolds for verifying each domain flow

Documentation

  • AI workflow: feature-development loop inside a toolkit-managed project
  • Target projects: scaffold, add a domain later, sync upstream drift
  • Agents: CLI flags, exit codes, and JSON output shapes
  • Docs index: every reference doc in this repo

Contributing

Portfolio project. Issues are welcome. Pull requests are accepted by invitation only.

License

MIT

Ver o original no GitHub

Bom para

Relacionados