Concepts & Guides¶ These guides explain the ideas behind Kaval.AI and how its pieces fit together. They complement the hands-on Tutorials — read a guide for the why, run a notebook for the how. Workflows The graph model Data types and validation Node types Context and interpolation YAML vs. WorkflowBuilder Streaming a run The WorkflowState Tools Three tool kinds, one interface Typed in, typed out Python tools REST tools MCP tools Introspection Agents Construction Running a prompt The four-step cycle Structured output The agent workflow node Observability What a run records Persistence and logging Local vs. production databases AgentService TaskLogger and fire-and-forget The backoffice UI Safety Typed I/O everywhere A safe expression language Bounded agents, explicit tools Deterministic testing