agent-architecture - 技术专题

《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：

Build AI agents from first principles using a local LLM - no frameworks, no cloud APIs, no hidden reasoning.

One agent. Every kind of work.

A 7×6 framework for agent architecture. 28 patterns, each placed at a coordinate, runnable Python code with verified engineering slices from Claude Code, Aider, OpenHands, DeerFlow. Companion to Designing AI Agents (Manning) by Jia Huang.

Chapter-by-chapter teardown of a production coding agent — agent loop, context engineering, compaction, permissions, MCP. 13 chapters, Chinese and English.

Engineering deterministic, production-grade systems around non-deterministic LLMs — FSM, durable execution, retries, DAGs, agent runtimes, model routing, edge inference, RAG, memory, multi-agent orchestration, security, and observability. 14 runnable proof-of-concept phases.

hermes-agent源码解析

The Architecture Review Agent is an open-source AI agent sample that reviews software architectures and generates interactive diagrams automatically.

Benchmarking the gap between AI agent hype and architecture. Three agent archetypes, 73-point performance spread, stress testing, network resilience, and ensemble coordination analysis with statistical validation.

Build production-ready AI agents in Claude Code from natural language. Analyzes your problem, researches best practices, and generates subagents, skills, hooks, commands & MCP configs.

🎙️ Voice-native document intelligence using Gemini, ElevenLabs STT/TTS, and Datadog observability — turning text documents into spoken conversations.

Universal autonomous agent framework with ReAct loop, multi-provider LLM routing, reasoning graph, and MCP integration, domain-agnostic for building specialized AI agents.

AI Agent engineering and quantitative finance knowledge base: source-level architecture analysis, Agent-to-Agent protocol research, runnable code experiments, industry insights. Obsidian vault, ready to clone.

An agent that compiles git maintenance chores into reusable programs, then decides whether to replay them by running checks rather than by guessing the task looks familiar

用钱学森工程控制论重塑 AI Agent 架构。前馈、积分控制、层级分离、自监控——让 Agent 真正懂你。—— Feedforward, integral control, hierarchical decomposition, and self-monitoring for a self-evolving agent that truly understands you.

《深入理解 DeepSeek Harness：一切皆插件的 Agent 架构》——源码级架构拆解科普书：37 个章节文件、PDF、Mermaid 图与写作规范

A comprehensive reference for building and organizing AI agent skills across Claude Code, KiloCode, OpenClaw, and OpenAI Codex. Covers SKILL.md specifications, folder structures, mode-scoped skills, agent orchestration, and self-hosted model configuration.

Reference implementation of a list-building path for GTM engineers: a plain-English campaign brief becomes a deduped, qualified, enriched, sequencer-ready contact list in seven stages, with swappable data providers.

A Recursive Ontological Framework for Cognitive Design, Neurodivergence Modeling, AI Co-Development, and "Structural AI"

OpenClaw source code architecture analysis of a production AI Agent gateway — Agent Loop, memory, tool sandbox, custom WebSocket protocol, 20+ IM adapters. 从 Agent Loop、记忆系统、工具沙箱到自研 WebSocket 协议 —— 拆解一个真正跑在生产环境中的多通道 AI Agent 网关是如何设计与实现的。