智能体团队编排
Agent Team Orchestration Skill
by amdf01-debug
安装
claude skill add --url github.com/openclaw/skills/tree/main/skills/amdf01-debug/sw-agent-team-orch文档
Trigger
Set up multi-agent teams with defined roles, task lifecycles, handoff protocols, and review workflows.
Trigger phrases: "multi-agent team", "agent orchestration", "set up agents", "task routing", "agent handoff", "agent coordination"
Process
- Define roles: What each agent specialises in
- Task lifecycle: inbox → spec → build → review → done
- Handoff protocol: How agents pass work between each other
- Quality gates: Review checkpoints before work moves forward
- Shared state: How agents share context and artifacts
Team Architecture Template
# Agent Team: [Name]
## Roles
### Manager Agent
- Routes incoming tasks to specialists
- Reviews completed work before delivery
- Escalates blocked tasks to human
- Model: [recommended model for this role]
### Specialist Agent: [Role Name]
- Handles: [task types]
- Outputs: [deliverable format]
- Quality bar: [minimum criteria]
- Model: [recommended model]
## Task Lifecycle
1. **Inbox**: New task arrives → Manager triages
2. **Assigned**: Manager routes to specialist with brief
3. **In Progress**: Specialist works, updates shared state
4. **Review**: Manager (or reviewer agent) checks output
5. **Revision**: If quality gate fails → back to specialist with notes
6. **Done**: Approved → delivered to requester
## Handoff Protocol
- Include: task description, context, acceptance criteria, deadline
- Never: assume context from previous tasks — always be explicit
- Format: structured JSON or markdown brief
## Quality Gates
- [ ] Output matches acceptance criteria
- [ ] No hallucinated data
- [ ] Formatting matches specification
- [ ] All links/references verified
- [ ] Spell-checked and proofread
Rules
- One task per agent at a time (focus > multitasking)
- Always include acceptance criteria in task briefs
- Shared state in files, not in agent memory (survives restarts)
- Model selection matters: use cheap models for bulk, expensive for judgment
相关 Skills
Claude接口
by anthropics
面向接入 Claude API、Anthropic SDK 或 Agent SDK 的开发场景，自动识别项目语言并给出对应示例与默认配置，快速搭建 LLM 应用。
✎ 想把Claude能力接进应用或智能体，用claude-api上手快、兼容Anthropic与Agent SDK，集成路径清晰又省心
RAG架构师
by alirezarezvani
聚焦生产级RAG系统设计与优化，覆盖文档切块、检索链路、索引构建、召回评估等关键环节，适合搭建可扩展、高准确率的知识库问答与检索增强应用。
✎ 面向RAG落地，把知识库、向量检索和生成链路系统串联起来，做架构设计时更清晰，也更少踩坑。
多智能体架构
by alirezarezvani
聚焦多智能体系统架构设计，梳理 Supervisor、Swarm、分层和 Pipeline 等模式，覆盖角色定义、通信协作与性能评估，适合规划稳健可扩展的 AI agent 编排方案。
✎ 帮你系统解决多智能体应用的架构设计与协同编排难题，适合构建复杂 AI 工作流，成熟度高、社区认可也很亮眼。
相关 MCP 服务
知识图谱记忆
编辑精选by Anthropic
Memory 是一个基于本地知识图谱的持久化记忆系统，让 AI 记住长期上下文。
✎ 帮 AI 和智能体补上“记不住”的短板，用本地知识图谱沉淀长期上下文，连续对话更聪明，数据也更可控。
顺序思维
编辑精选by Anthropic
Sequential Thinking 是让 AI 通过动态思维链解决复杂问题的参考服务器。
✎ 这个服务器展示了如何让 Claude 像人类一样逐步推理，适合开发者学习 MCP 的思维链实现。但注意它只是个参考示例，别指望直接用在生产环境里。
PraisonAI
编辑精选by mervinpraison
PraisonAI 是一个支持自反思和多 LLM 的低代码 AI 智能体框架。
✎ 如果你需要快速搭建一个能 24/7 运行的 AI 智能体团队来处理复杂任务（比如自动研究或代码生成），PraisonAI 的低代码设计和多平台集成（如 Telegram）让它上手极快。但作为非官方项目，它的生态成熟度可能不如 LangChain 等主流框架，适合愿意尝鲜的开发者。