iron-manus-mcp

概述

Iron Manus MCP是一个复杂的MCP服务器，专为AI工作流编排而设计。它实现了8阶段工作流，包含JARVIS工作流控制器和具有SSRF保护的APITaskAgent等专业化工具。服务器引入了许多后来成为行业标准的创新模式，包括将待办事项作为子代理分派队列、基于阶段的工具访问权限控制和结构化的规划阶段。

试试问 AI

装完之后，这里有 6 个你可以让 AI 做的事：

• JARVIS

  8-phase workflow controller for orchestration

  试试:Start a new 8-phase workflow试试:Initialize a project using the Iron Manus workflow
• APITaskAgent

  API discovery and fetching with SSRF protection

  试试:Fetch data from GitHub API试试:Discover and call a REST API endpoint
• PythonComputationalTool

  Python execution for data analysis

  试试:Analyze this dataset with Python试试:Perform statistical calculations with Python
• IronManusStateGraph

  Session state management

  试试:Save current workflow state试试:Restore workflow from previous state
• SlideGenerator

  HTML slide generation

  试试:Create a presentation about project updates试试:Generate slides for the quarterly review
• HealthCheck

  Runtime diagnostics

  试试:Check system health试试:Verify all components are running

可对比工具

安装

git clone https://github.com/dnnyngyen/iron-manus-mcp
cd iron-manus-mcp
npm install
npm run build
npm start

docker build -t iron-manus-mcp .
docker run -d --name iron-manus-mcp iron-manus-mcp

claude mcp add iron-manus-mcp node dist/index.js

{
  "mcpServers": {
    "iron-manus-mcp": {
      "command": "node",
      "args": ["path/to/iron-manus-mcp/dist/index.js"]
    }
  }
}

FAQ

8阶段工作流是什么？

Iron Manus MCP实现了8阶段工作流（INIT → QUERY → ENHANCE → KNOWLEDGE → PLAN → EXECUTE → VERIFY → DONE），强制执行规划阶段，并结构化代理交互。

如何将待办事项用作子代理分派？

待办事项充当任务分派机制，每个待办事项可以指定代理类型（TaskAgent、SubAgent、DirectExecution）和配置，实现结构化任务委派。

iron-manus-mcp 对比