agent harness engineering
An MIT-licensed educational repository that teaches building a nano Claude Code-like agent harness from scratch in Python.
Same problem, same approach. Swapping one for another is a config change, not a rewrite.
Choose for a comprehensive Apache-2.0 Chinese book on agent design with chapter code, instead of a minimal MIT-licensed tutorial.
《深入理解 AI Agent：设计原理与工程实践》（李博杰 著）开源主仓库：全书正文、编译版 PDF 与按章配套代码
Choose when you prefer a TypeScript-based MIT tutorial from 0 to 1 over the Python-based seed.
Harness engineering beginner tutorial, from 0 to 1
Choose when you need a much deeper Chinese-language, 15-chapter treatment culminating in building your own harness.
《御舆：解码 Agent Harness》42万字拆解 AI Agent 的Harness骨架与神经 —— Claude Code 架构深度剖析，15 章从对话循环到构建你自己的 Agent Harness。在线阅读网站：
Choose for a source-level analysis of the real Claude Code if you want to understand existing internals rather than build a simplified version.
Deep dive into Claude Code internals — architecture, agent loop, context engineering, and more. / 深入解析 Claude Code 源码：架构、Agent 循环、上下文工程、工具系统等
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Choose for a broader HTML guide covering LangGraph and RAG when you need survey-level content rather than building a harness from scratch.
https://adongwanai.github.io/AgentGuide | AI Agent开发指南 | LangGraph实战 | 高级RAG | 转行大模型 | 大模型面试 | 算法工程师 | 面试题库 | 强化学习｜数据合成
Choose for a runnable textbook/library of 35 architectures (MIT) when you need a multi-provider survey instead of a single minimal build.
35 production-grade agentic AI architectures (Reflexion, LATS, GraphRAG, MemGPT, Voyager, BrowserAgent, ...) — a Python library and runnable textbook with multi-provider LLM support and a 17-task benchmark leaderboard.
Choose for a MIT-licensed first-principles reconstruction of Claude's architecture if you want a full reference implementation to study.
A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.
Choose for a production multi-agent framework with extensive features and ecosystem when you don't want to hand-roll a minimal harness.
A programming framework for agentic AI
Does the same job, but ships as an app. Useful to a person, not swappable into a codebase.
Choose when you need a ready-to-use macOS offline harness with memory and identity as an end-user app, instead of code to build your own.
Own your AI. The native macOS harness for AI agents -- any model, persistent memory, autonomous execution, cryptographic identity. Built in Swift. Fully offline. Open source.
These projects were analysed and named learn-claude-code among their alternatives. The relationship is not symmetric — how learn-claude-code rates them is a separate judgement, made when learn-claude-code is analysed in its own right.
calls learn-claude-code “choose for a project-based from-0-to-1 harness build focused on a Claude Code-like nano agent, directly comparable to the seed”
Harness engineering beginner tutorial, from 0 to 1
calls learn-claude-code “Choose if you prefer Python and a minimal nano harness built from 0 to 1 instead of TypeScript checkpoints.”
《动手学 Pi》：沿 15 个真实 checkpoint 从零构建 Pi-style Agent
calls learn-claude-code “Hands-on coding tutorial that builds a minimal harness from scratch; choose it if you learn by building rather than reading.”
《御舆：解码 Agent Harness》42万字拆解 AI Agent 的Harness骨架与神经 —— Claude Code 架构深度剖析，15 章从对话循环到构建你自己的 Agent Harness。在线阅读网站：
calls learn-claude-code “Step-by-step build of a Claude Code-like harness, scoped to that one product.”
《深入理解 AI Agent：设计原理与工程实践》（李博杰 著）开源主仓库：全书正文、编译版 PDF 与按章配套代码
calls learn-claude-code “Choose when you want a tiny, educational agent harness you can read and modify, for simple coding tasks.”
Trae Agent is an LLM-based agent for general purpose software engineering tasks.
calls learn-claude-code “Choose when you want to build a nano claude-code-like harness from scratch.”
https://adongwanai.github.io/AgentGuide | AI Agent开发指南 | LangGraph实战 | 高级RAG | 转行大模型 | 大模型面试 | 算法工程师 | 面试题库 | 强化学习｜数据合成
calls learn-claude-code “Choose it for a transparent bash-based harness when you need a simple loop without sandboxing or event logging.”
Apache Maka (Incubating) is a local-first AI agent workspace. Model messages, tool calls, tool results, permission decisions, and termination events are recorded as an append-only log.