The Agentic Wiki is a resource dedicated to documenting, explaining, and advancing the understanding of agentic AI design patterns. As large language models move from simple prompt-response interactions to autonomous, goal-directed systems, engineers need clear architectural guidance for building reliable agents.
Our Mission
The field of agentic AI is evolving rapidly, but much of the best knowledge is scattered across research papers, blog posts, and tribal expertise. This wiki aims to collect, organize, and present that knowledge in a clear, practical format. Our goal is not to promote any single framework or vendor, but to explain the underlying patterns and principles that make agentic systems work.
Foundation
The core pattern catalog is grounded in practical engineering concerns: reliability, observability, evaluation, and safe tool use. We combine foundational workflow patterns with advanced topics like memory architectures, guardrails, and composition strategies so teams can make context-aware design decisions.
Contributing
We do not yet have a clearly defined contribution model that balances openness with quality in a world of rapidly increasing AI-generated content. We are actively working toward a community-friendly and sustainable approach.
In the meantime, ideas, feedback, and suggestions are very welcome. If you have thoughts on what a good contribution model should look like, please share them in the project repository.
Staying Current
Agentic AI is a fast-moving field. We aim to keep this wiki up to date as new patterns emerge, existing frameworks evolve, and the community develops better practices. If you notice something outdated or missing, please let us know.