Skip to contents

Builds the linear graph equivalent of agent_pipeline() (node i is agent i, each feeding the next). The original agent_pipeline() keeps its own simple implementation; this exists to show the graph can express it and to let a linear pipeline gain checkpointing when wanted. The first agent reads state$input; the nodes are named stage_1, stage_2, and so on, and each writes its reply into the state under its own name.

Usage

workflow_from_pipeline(agents)

Arguments

agents

A list of Agents, in order (a single Agent is accepted).

Value

An agent_workflow.