# Agent Runtime Testkit
`agent-runtime-testkit` provides deterministic fakes, fixtures, and assertions
for testing Agent definitions and Kernel-facing contracts without a live model,
workspace, conversation service, or Agent Infra deployment.
Use it in `dev-dependencies` beside `agent-runtime-sdk`. Scripted sessions keep
requests observable and return responses in a deterministic order, making tool
loops and failure cases reproducible.
See the [API documentation](https://docs.rs/agent-runtime-testkit).