Practical overview
When it fits
Use it when an agent project needs more than a flat list of patterns and the team wants to relate an agent’s cognitive function to its runtime topology before choosing an implementation approach.
Who it is for
- AI-agent developers designing or refactoring agent workflows.
- Technical leads and reviewers comparing architecture patterns.
- Educators and learners studying reusable agent-design structures.
- Teams creating internal standards for agent implementation reviews.
Documented uses
- Select an agent architecture using a structured pattern matrix.
- Study reference implementations for recurring agentic workflow shapes.
- Adapt documented patterns into components for a custom agent.
- Create a shared vocabulary and review process for agent designs.
- Compare documented engineering examples associated with established agent projects.
Requirements & limits
The repository is public and lists an MIT license. It is available as a web-based GitHub repository with English and Chinese documentation.
- The matrix currently groups patterns across seven cognitive functions and six execution topologies.
- Pattern entries differ in implementation status: the public matrix marks some cells as runnable code and others as scaffolded.
- The repository combines documentation and code, allowing a design concept to be examined alongside a reference implementation where available.
- Its scope is architecture guidance and reference material rather than a hosted agent runtime.
- A separate companion repository is identified for the chapter-by-chapter Argus example, while this repository serves as the standalone pattern catalogue.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Map an agent workflow to a documented pattern
Use the repository’s matrix to select an architectural pattern by cognitive function and execution topology.
Before you start
- Access to the public Agent Design Patterns repository.
- A defined agent task and its intended cognitive function.
- An understanding of the desired runtime topology, such as chain, route, parallel, loop, hierarchy or orchestrate.
Steps
Describe what the agent must do in cognitive terms, such as perceive, remember, reason, act, reflect, collaborate or govern.
Choose the execution topology that matches how the work should be laid out at runtime.
Locate the corresponding matrix coordinate and open the linked pattern folder and README.
Review the available code or scaffold status before adapting the pattern to the agent workflow.
Expected outcome
A pattern-based starting point for structuring or reviewing an AI-agent architecture.
Guide limitations
- The repository matrix includes both runnable-code cells and scaffolded cells, so implementation material varies by pattern.
- The guide describes repository-based design selection and does not establish that an adapted implementation will run in a particular environment.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 3 October 2026
- Original website title
- huangjia2019/agent-design-patterns · Official homepage title · retained in its source language