Agentics Patterns
Agentic ideas you can run and inspect.
A collection of runnable workflows that makes agentic patterns concrete enough to study, compare, and test.
A pattern should be inspectable.
Agentic systems are easier to discuss than to reason about precisely. I built Agentics Patterns to turn workflow ideas into concrete examples with inputs, flow definitions, and expected results.
Make the decisions visible.
- Describe the workflow so a reader can follow where a decision happens.
- Keep input and expected output explicit, so examples can be compared.
- Treat contracts and checks as part of the workflow design.
From a flow to a testable example.
The examples connect flow definitions with sample inputs and expected outputs. That gives developers a starting point for understanding the boundaries and adapting a pattern to their own problem.
Useful examples still need judgment.
A runnable pattern is a learning tool. Applying it in production still requires choosing the right boundaries, permissions, evaluation criteria, and failure behavior for the actual use case.
Read the flow. Run the example.
The repository contains the workflows and their supporting artifacts. Use it to explore the patterns, examine the contracts, and build your own understanding of agentic execution.
Have a complex challenge?
Let’s make sense of it.
I’m interested in conversations about engineering leadership, practical AI, open source, and technology that matters.