Designing Ai Agents

Designing AI Agents — Source Code

**[*Designing AI Agents*](https://hubs.la/Q04hCsH10)** — the design-pattern catalogue for production AI agents. (Manning)

This repository is the official source code for the book. It contains two tracks living side by side inside each chapter:

• argus/ — Argus, the running-example PR-review agent, evolves cumulatively from Ch2 to Ch10. Each chapter adds one cognitive module and ships a self-contained snapshot that imports as from argus import ... so the reader runs the chapter's Argus directly: python -m argus.cli --project .
• patterns/ — independent pattern demos for everything the chapter introduces. Not every pattern integrates into Argus. Those patterns live here as runnable references.

中文版：[README.zh-CN.md](README.zh-CN.md)

Need a separate **pattern-only catalog organized by the two-axis matrix instead of by chapter?** See [**huangjia2019/agent-design-patterns**](https://github.com/huangjia2019/agent-design-patterns) (ADPS asset, not part of the book).

---

Argus evolution at a glance

Chapter	Argus +=	Module added	What you can run
Ch2 architecture	single-pass PRA loop	argus/core.py	(seed; one LLM call)
Ch3 perception	+ gather & triage code context	argus/perception.py	python -m argus.cli
Ch4 memory	+ cross-session memory (RAG)	argus/memory.py	... --project myapp
Ch5 reasoning	+ complexity-routed CoT	argus/reasoning.py	tier=simple/moderate/complex
Ch6 action	+ tool dispatch + Guardrail Sandwich	argus/action.py	lint / test / fix_apply
Ch7 reflection	+ critic loop + skill library