compose-agents

What problem does it solve?

It helps you design reliable multi-agent workflows by coordinating when one agent runs after another, when multiple agents run simultaneously, and when an iterative review/repair loop should continue.

Core Features & Use Cases

• SequentialAgent: Chain agents so the next agent uses the previous agent’s final output as context.
• ParallelAgent: Run multiple agents concurrently with branch isolation, then combine results downstream.
• LoopAgent: Repeat a set of agents until a stop condition is met, supporting fixed iteration limits or custom continuation logic.
• Transfer routing: Hand off work between specialized sub-agents using transfer tools, enabling triage and delegation patterns.

Quick Start

Use compose-agents to build a pipeline that runs research in parallel, then synthesizes the results sequentially, and finally reviews the draft in an approval loop.