Chapter 6: Multi-Agent and Advanced Patterns
April 13, 2026 · View on GitHub
Welcome to Chapter 6: Multi-Agent and Advanced Patterns. In this part of Strands Agents Tutorial: Model-Driven Agent Systems with Native MCP Support, you will build an intuitive mental model first, then move into concrete implementation details and practical production tradeoffs.
This chapter explores advanced usage beyond basic single-agent workflows.
Learning Goals
- compose multiple agents for specialized workflows
- evaluate when to add streaming or autonomous patterns
- balance complexity with maintainability
- define escalation paths for human oversight
Advanced Pattern Areas
- multi-agent decomposition by domain responsibility
- bidirectional streaming experiences for real-time interactions
- hybrid tool + MCP designs for broader capability coverage
Source References
Summary
You now have a roadmap for scaling Strands workflows without losing architectural control.
Next: Chapter 7: Deployment and Production Operations
Source Code Walkthrough
Use the following upstream sources to verify multi-agent and advanced pattern details while reading this chapter:
src/strands/multiagent/— the multi-agent coordination module containing swarm patterns, agent-as-tool composition, and pipeline orchestration primitives.src/strands/multiagent/swarm.py— the swarm implementation that enables parallel agent execution with shared context and result aggregation.
Suggested trace strategy:
- review
src/strands/multiagent/to understand the available coordination patterns (swarm, pipeline, agent-as-tool) - trace how
swarm.pymanages concurrent agent execution and merges outputs - check the agent-as-tool pattern to understand how one agent can be registered as a callable tool for another agent
How These Components Connect
flowchart LR
A[Orchestrator agent] --> B[multiagent/ coordination]
B --> C[Swarm: parallel sub-agents]
B --> D[Pipeline: sequential agent chain]
C --> E[Aggregated result]
D --> E