ac-init

Scan project structure and generate CLAUDE.md documentation for multi-module repositories.

6|2|Updated Jan 9, 2026
One-click install
npx skills add https://github.com/yz0812/ai-dev-toolkit --skill ac-init
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ac-init
Source: https://github.com/yz0812/ai-dev-toolkit/tree/main/skills/ac-init
Command: npx skills add https://github.com/yz0812/ai-dev-toolkit --skill ac-init

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

帮助开发团队快速梳理并生成 CLAUDE.md 的上下文文档,通过主线程拉取项目信息并可选地让 subagents 同步扫描以完善文档。

Core Features & Use Cases

  • 主线程负责时间戳、决定是否启用 subagents、汇总扫描结果、写入根级与模块级 CLAUDE.md
  • 当项目较大或需要并行梳理时,使用 subagents 进行只读扫描以收集模块职责、关键文件和依赖关系
  • 如指定 --modules,生成模块级 CLAUDE.md,组织模块结构与接口关系

Quick Start

在工作目录中运行 /ac-init [项目简述] [--modules] 来生成根级 CLAUDE.md,并按需要生成模块级 CLAUDE.md。

Frequently Asked Questions about ac-init

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate CLAUDE.md context documentation for a multi-module repository?

To generate CLAUDE.md context documentation, run the /ac-init command with an optional project description. It scans the project structure, module responsibilities, key files, and dependencies to produce root-level and module-level documentation artifacts.

How does automated project structure scanning work for generating context files?

Automated project structure scanning works by using the Claude main thread to pull project information and optional subagents to perform read-only scans. This parallel scanning collects module responsibilities, key files, and dependency relationships to compile comprehensive context files.

When do I need to use the --modules flag for module documentation generation?

You need the --modules flag when generating module-level CLAUDE.md files for multi-module repositories. It organizes module structures and interface relationships, creating specific documentation artifacts alongside the root-level context file.

Does the project structure scanner skip build and cache directories automatically?

Yes, the project structure scanner enforces ignore rules to automatically skip build and cache directories. This ensures the generated CLAUDE.md context documentation only includes relevant source files and module structures.

Can I use subagents to scan large project structures in parallel?

Yes, you can use subagents to scan large project structures in parallel. The main thread decides whether to enable subagents, which perform read-only scans to collect module responsibilities and dependency relationships for comprehensive documentation.

What is the best way to document module responsibilities and key files in a large codebase?

The best way to document module responsibilities is using automated scanning to identify project structure and generate CLAUDE.md context files. It captures module responsibilities, key files, and dependencies, providing a clear overview for multi-module repositories.