ZhangHanDong/harness-engineering-from-cc-to-ai-coding resource

Health v2 · maintenance only

Flags: no_releases young

Adoption not part of the score

1492 stars · 272 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

An open-source Chinese technical book (nicknamed 'The Horse Book') analyzing Claude Code's v2.1.88 source code to extract AI coding agent architecture patterns, context strategies, permission systems, and production practices. It is built with mdBook and published on GitHub Pages, with an English translation in progress.

Use cases

• learn how Claude Code works internally
• understand AI coding agent architecture patterns
• study agent loop and tool orchestration design
• learn context management and prompt caching strategies for LLM apps
• design permission and safety systems for AI agents
• build a custom AI coding agent
• optimize LLM application cost and latency

When to choose

• you are building AI coding agents or agent frameworks and want real-world engineering patterns
• you want a systematic, book-style deep dive into Claude Code's implementation rather than official docs
• you are a security engineer studying AI agent permission models and defense in depth
• you want to learn prompt caching, compaction, and token budgeting techniques

When to avoid

• you need official Anthropic documentation or product usage guides
• you want a software tool or library to install rather than reading material
• you need the English version fully complete (translation is still in progress)
• you require legally sanctioned source material - the book is based on reverse-engineered release packages

Facets

learning-resource · maturity active

documentation developer-tools agent-framework prompt-engineering artificial-intelligence large-language-models developer-tools tutorials cli cross-platform book claude-code mdbook reverse-engineering ai-coding agent-architecture context-management prompt-caching chinese ai-agents web-server

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ZhangHanDong/harness-engineering-from-cc-to-ai-coding")