Agent Design Patterns
A reference implementation and bilingual documentation of 27 core agent-architecture patterns, each placed on a two-axis map of cognitive function against execution topology, as a companion to the book Designing AI Agents.
Agent Design Patterns organizes agent-architecture patterns onto a coordinate system rather than a flat list, on the premise that a pattern only helps once you know where in an agent's design a given problem actually sits. The two axes, cognitive function and execution topology, come from an accompanying paper, and each of the 27 core patterns and its extensions ships with runnable Python code and engineering examples drawn from Claude Code, Aider, OpenHands and DeerFlow.
The repository holds both the reference code and editable documentation for the patterns in English and Chinese, with published articles linking back to the exact source commit they describe. It is the standalone pattern catalog; a separate companion repository grows one evolving codebase chapter by chapter to follow the book's narrative.