将AI文物(技能,规则,命令)分配到整个开发团队的端到端管道——完全由AI协助建造.

这个寄存器伴有**“Od shazeu do AI-Native Information”** webinar,演示了如何从"50个开发者,50个不同的提示"到一个标准化的,版本控制的有私人注册和CI/CD的AI工具包.

∮我们所构建的∮

• 步骤 * * 构成部分 * 说明 * * |------------. +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

---

3 ** 私人登记** 通过Terraform提供AWS代码Artiffact——管理式npm登记,模式为双repo . ** CI/CD管道 ** 与 OIDC 认证的 GitHub 动作工作流程,用于自动验证和发布 * **

• • 5 * ** 团队安装 ** * * 为任何项目类型的任何开发者设置一个命令 * *

□ 建筑

二安装模式

• ** JS/TS项目**:npm install @10xdevs/ai-toolkit——邮站建立连锁关系

AI-Native Infrastructure

End-to-end pipeline for distributing AI artifacts (skills, rules, commands) across a development team — built entirely with AI assistance.

This repository accompanies the “Od chaosu do AI-Native Infrastructure” webinar, demonstrating how to go from “50 developers, 50 different prompts” to a standardized, version-controlled AI toolkit with private registry and CI/CD.

What We Build

Architecture

Team Conventions          Private Registry (AWS CodeArtifact)
       │                         ▲            │
       ▼                         │            ▼
  Agent Skill ──► Extension Pack ──► npm publish    npm install / npx install
  (SKILL.md)     (pack.yaml)     │                         │
                                 │                         ▼
                            CI/CD Gate              .claude/skills/
                         (GitHub Actions)           .claude/commands/

Two Installation Modes

• JS/TS projects: npm install @10xdevs/ai-toolkit — postinstall creates symlinks
• Any project (Python, Go, Rust…): npx @10xdevs/ai-toolkit install — copies files, no package.json needed

Both modes install into the project’s .claude/ directory and track state via a manifest for clean uninstall.

Repository Structure

context/
  conventions.md        # Team engineering conventions (input for skill creation)
  spec-skill.md         # Requirements: Agent Skill
  spec-pack.md          # Requirements: Extension Pack
  spec-terraform.md     # Requirements: Terraform registry
  spec-cicd.md          # Requirements: CI/CD pipeline
packages/ai-toolkit/    # The Extension Pack (npm package)
  pack.yaml             # Pack manifest
  skills/code-review/   # Code review Agent Skill
  install.js            # Dual-mode installer (symlink/copy)
  bin/cli.js            # CLI for npx usage
skills/                 # Project-level skills
terraform/              # IaC for AWS CodeArtifact registry
.github/workflows/      # CI/CD pipeline

Git Checkpoints

The repo has tagged branches for each stage of the build, so you can jump to any point: