Back to Details

compose

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

MANDATORY PREPARATION

强制准备工作

Invoke {{command_prefix}}agent-workflow — it contains workflow principles, anti-patterns, and the Context Gathering Protocol. Follow the protocol before proceeding — if no workflow context exists yet, you MUST run {{command_prefix}}teach-maestro first. Consult the agent-architecture reference in the agent-workflow skill for topology patterns and when multi-agent is justified.
Design a multi-agent system. But first — are you sure you need one?
调用 {{command_prefix}}agent-workflow —— 它包含工作流原则、反模式，以及上下文收集协议。继续操作前请遵循该协议——如果还不存在工作流上下文，你必须首先运行 {{command_prefix}}teach-maestro。 请查阅agent-workflow技能中的agent-architecture参考，了解拓扑模式以及适合使用多Agent的场景。
设计一个多Agent系统。但首先——你确定真的需要多Agent系统吗？

Step 1: Pre-Composition Check

步骤1：组合前检查

Answer these before proceeding:
  1. Has a single agent been tried and failed? (If no, try single agent first)
  2. What specific limitation requires multiple agents? (If you can't name it, you don't need multi-agent)
  3. Is the cost/latency increase justified? (Multi-agent = 2-10x cost and latency)
If you can't articulate a specific limitation, use {{command_prefix}}amplify on the single agent instead.
继续操作前请先回答以下问题：
  1. 是否已经尝试过单个Agent但失败了？（如果没有，请先尝试使用单个Agent）
  2. 需要使用多个Agent的具体限制是什么？（如果你说不出来，说明你不需要多Agent）
  3. 成本/延迟的提升是否合理？（多Agent = 2-10倍的成本和延迟）
如果你无法明确说明具体的限制，请改为对单个Agent使用 {{command_prefix}}amplify。

Step 2: Design the Topology

步骤2：设计拓扑结构

Choose the right architecture pattern (consult the agent-architecture reference in the agent-workflow skill):
For each agent in the system, define:
markdown
undefined
选择合适的架构模式（请查阅agent-workflow技能中的agent-architecture参考）：
为系统中的每个Agent定义如下信息：
markdown
undefined

Agent: [Name]

Agent: [Name]

Role: [One sentence] Responsibilities: [What it does] Boundaries: [What it does NOT do] Tools: [List of tools this agent has access to] Input: [What it receives] Output: [What it produces]
undefined
Role: [One sentence] Responsibilities: [What it does] Boundaries: [What it does NOT do] Tools: [List of tools this agent has access to] Input: [What it receives] Output: [What it produces]
undefined

Step 3: Design Handoffs

步骤3：设计交接流程

For each agent-to-agent connection:
markdown
undefined
为每一个Agent间的连接定义如下信息：
markdown
undefined

Handoff: [Agent A] → [Agent B]

Handoff: [Agent A] → [Agent B]

Trigger: [When does A hand off to B?] Payload: [What data is passed?] Expected response: [What does A expect back?] Timeout: [How long to wait?] Failure handling: [What if B fails?]
undefined
Trigger: [When does A hand off to B?] Payload: [What data is passed?] Expected response: [What does A expect back?] Timeout: [How long to wait?] Failure handling: [What if B fails?]
undefined

Step 4: Design the Supervisor

步骤4：设计监管器

Every multi-agent system needs a supervisor:
  • Monitors agent health and performance
  • Routes tasks to appropriate agents
  • Handles failures and escalation
  • Enforces global constraints (budget, time, quality)
每个多Agent系统都需要一个监管器：
  • 监控Agent的运行状态和性能
  • 将任务路由给合适的Agent
  • 处理故障和升级流程
  • 执行全局约束（预算、时间、质量）

Composition Checklist

组合检查清单

  • Each agent has a clear, non-overlapping role
  • Handoff protocols are defined for every connection
  • A supervisor pattern is in place
  • Cost/latency budget accounts for all agents
  • Failure modes are handled at every handoff point
  • The system can be understood by reading the topology diagram
  • 每个Agent都有清晰、不重叠的角色
  • 所有连接都定义了交接协议
  • 已配置监管器模式
  • 成本/延迟预算覆盖所有Agent的消耗
  • 每个交接节点都处理了故障模式
  • 通过阅读拓扑图就可以理解整个系统

Recommended Next Step

推荐后续步骤

After composition, run 
{{command_prefix}}fortify
to add error handling at every handoff, then 
{{command_prefix}}evaluate
to test the multi-agent system end-to-end.
NEVER:
  • Build multi-agent for a problem a single agent can handle
  • Create agents with overlapping responsibilities
  • Skip the supervisor (autonomous swarms are unpredictable)
  • Pass full context between all agents (pass only what's needed)
  • Compose without defining handoff protocols
完成组合后，运行 
{{command_prefix}}fortify
为每个交接环节添加错误处理，然后运行 
{{command_prefix}}evaluate
对多Agent系统进行端到端测试。
绝对禁止：
  • 为单个Agent就能解决的问题构建多Agent系统
  • 创建职责重叠的Agent
  • 省略监管器（自主集群的运行结果不可预测）
  • 在所有Agent之间传递完整上下文（仅传递需要的内容）
  • 未定义交接协议就进行系统组合