layer0
Protocol traits for composable agentic AI systems
Overview
layer0 defines the foundational protocol traits that the entire neuron workspace builds on. It
contains no implementations — only the contracts that every agentic component must satisfy.
The traits here form the 6-layer model:
| Layer | Trait(s) | Responsibility |
|---|---|---|
| 0 | Operator, Effect |
Invoke operators, emit structured effects |
| 1 | StateStore |
Persistent key-value memory |
| 2 | Environment, CredentialRef |
Credential injection + env config |
| 3 | Orchestrator, WorkflowHandle |
Coordinate multi-operator workflows |
| 4 | Hook, HookRegistry |
Pre/post lifecycle middleware |
| 5 | Observable |
Structured observability events |
Usage
[]
= "0.4"
Test utilities
[]
= { = "0.4", = ["test-utils"] }
The test-utils feature exports stub implementations (StubOperator, StubStateStore, etc.)
useful for testing downstream crates without pulling in full implementations.
Part of the neuron workspace
neuron is a composable async agentic AI framework for Rust. See the book for architecture and guides.