agent-design 开源项目推荐（61 个）

huangjia2019 /agent-design-patterns

一个7×6代理架构框架。28个模式，每个配置在一个坐标上，可运行的Python代码，并包含Claude Code、Aider、OpenHands、DeerFlow等经过验证的工程切片。《设计人工智能代理指南》（Manning），作者：贾黄。

kangarooking /system-prompt-skills

从 165 个顶级 AI 产品系统提示词中蒸馏出的 15 个可执行 Agent 技能

OPBR /build-claude-code

从0到1构建 Claude code (学习版)

Saint1010-arch /agent-design-system

🦞 Production-grade AI Agent design system — 龙虾教练：产品级 AI Agent 设计体系

Khrylx /Transform2Act

[ICLR 2022 Oral] Official PyTorch Implementation of "Transform2Act: Learning a Transform-and-Control Policy for Efficient Agent Design".

AISmithLab /HumanStudy-Bench

HumanStudy-Bench: Towards AI Agent Design for Participant Simulation

AgriciDaniel /hive-mind

How to design AI agents that work as one: nine parts of an agent and its team, checked against 85 sources. Free guide in Markdown and PDF, with the drawings.

galyarderlabs /galyarder-design

Experimental design engine for AI coding agents. Derived from open-design.

SarutobiSasuke8 /Meta-Agent-OS

A Markdown-based operating system for diagnosing, designing, costing, building, and operating multi-agent AI systems. Methodology upstream of code.

HumanStudy-Hub /HumanStudy-Bench

HumanStudy-Bench: Community Edition — Standardized human study replays for AI agent evaluation

dadwadw233 /claude-code-philosophy

Vendor-neutral agent design skill inspired by Claude Code-class systems.

tjennychen /writing-system-prompts

Claude Code skill for writing system prompts — 14 sections, 12 common mistakes, 4 prompt types

MedocMay /ai-native-builder-consultant-skills

Consultant-grade skills, workflow, and templates for designing, evaluating, launching, and iterating AI Native products.

Mutlaq124 /wargame_multi-agent

Architected hierarchical multi-agent wargame simulation with strategic planning, tactical agents orchestration, real-time spatial reasoning, and adaptive memory module

SuperLogicAI /production-agent-design

Architect high-reliability, logic-driven AI agents for production.

VishApp /crewaimaster

CrewAIMaster transforms any task description into a fully functional, production-ready multi-agent system, making advanced AI orchestration accessible to everyone.

lilCnid /subagent-composition

Moved into OpenCnid/dovetail — this skill now ships in that pack at skills/subagent-composition/. Archived.

nraford7 /agent-studio-v2

MOVED to github.com/nraford7/agent-studio (archived, frozen at Aug 2026)

NachaFromMars /superbuild

Agentic engineering knowledge + RPI process (Research → Plan → Implement). 5 design patterns. Reference files per situation: context-engineering, orchestration, agent-design, memory-patterns.

Little-Cyrene /raising-an-ai

一个人格型 agent 的四天工程记录：能力边界、上下文治理、记忆载体、拒绝权与中断权（不是教程，是判断和坑）