multi-agent-orchestration
Installation
SKILL.md
Multi-Agent Orchestration
Design coordinated multi-agent systems using battle-tested orchestration patterns. This skill provides architectural guidance for moving from single-agent to multi-agent systems, choosing coordination patterns, and implementing production-grade orchestration layers.
When NOT to Use Multi-Agent
Start with the simplest approach that reliably meets requirements:
| Level | When to Use | Example |
|---|---|---|
| Direct model call | Single-step tasks (classification, translation, summarization) | "Summarize this article" |
| Single agent + tools | Varied queries within one domain requiring dynamic tool use | Order lookup with API calls |
| Multi-agent | Cross-domain problems, security boundaries, or parallel specialization | Financial underwriting (data extraction, scoring, compliance, risk) |
Rule: Only add multi-agent complexity when a single agent can't reliably handle the task due to:
- Prompt complexity or tool overload
- Security/permission isolation requirements
- Parallelization benefits that outweigh coordination overhead