dadiaomengmeimei/claude-code-sourcemap-learning-notebook

README preview

# Claude Code Sourcemap Learning Notebook

[![GitHub](https://img.shields.io/badge/GitHub-dadiaomengmeimei-blue?logo=github)](https://github.com/dadiaomengmeimei/claude-code-sourcemap-learning-notebook)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)

> 🌐 [中文版 / Chinese Version](README_CN.md)

**Reverse-engineering Claude Code's 512K+ lines of TypeScript — extracting the architecture, design decisions, and 11 transferable patterns that make a production AI Agent work.**

Not just a code walkthrough. An architectural analysis that explains *why* each design choice was made, with alternatives compared.

## What You'll Learn

- 🔄 **Query Loop** — Streaming tool execution, read-write lock concurrency, 3-layer AbortController cascade
- 🛠️ **Tool System** — `buildTool()` factory + Zod schema as prompt, 40+ tools with 18+ feature flags
- 🔒 **5-Layer Security** — Permission rules → mode → tool checks → path safety → macOS Seatbelt sandbox
- 🤖 **Multi-Agent** — Coordinator orchestration, default isolation with explicit sharing, cache-friendly forking
- 🔌 **MCP & Skills** — 3-layer skill architecture, bundled skill lazy extraction, `/simplify` 3-agent parallel review
- 🧠 **Prompt Engineering** — 7 categories with 40+ prompt files, static/dynamic separation for cache optimization
- 🎙️ **Voice & Buddy** — Record-first-connect-later latency elimination, deterministic generation

View full repository on GitHub →