Agent 开发框架

TeamHero|面向 Claude Code 的多智能体编排与任务管理平台

TeamHero

TeamHero 是一个自托管的多智能体编排平台,用任务计划、审批、交付物追踪和持久记忆来管理 AI 智能体团队。

获取与使用

需要部署

按官方资料自行安装或部署

使用难度待核验
国内可用性待核验

2026-11-09 更新 · 本站尚未实测

项目介绍

Agent 开发框架 · 2026-11-09 更新 · 本站尚未实测

TeamHero 是一个自托管的多智能体编排平台,仓库自述为“Manage AI agent teams with real project management discipline”。它的出发点是:管理 AI 智能体不应只是把它们启动起来碰运气,而应像管理真人团队一样有计划、有审批、有交付物追踪。

README 说明,作者先做了商业产品 Kapow(gokapow.com),随后将核心架构以 MIT 许可证开源为 TeamHero。项目主要面向使用 Claude Code、希望围绕智能体建立结构化管理流程的开发者,需要熟悉 CLI 与 Node.js。

需要注意:TeamHero 本身是编排平台,不是单个可直接导入客户端运行的 Agent 配置,也不是 MCP 服务器。

核心能力

点击能力名称查看输入、输出和原文依据。

  • Dashboard:通过 Web UI 一览管理整个智能体团队。
  • Command Center:在集成终端中与编排器(orchestrator)智能体对话。
  • 任务系统:包含计划、版本、审批和交付物追踪的完整生命周期。
  • 智能体记忆:每个智能体拥有跨会话的短期与长期持久记忆。
  • 知识库:将交付物提升为可检索、分类的库。
  • Skills:可选集成,README 举例为浏览器自动化与 GitHub。
  • 冲突预防:通过文件范围声明协调改动(不是操作系统访问隔离,不能保证完全防止覆盖)。

实际应用场景

适用场景

适合多个智能体在同一项目中协作、需要先计划再执行并追踪交付物的开发者团队;也适合希望以自托管方式研究规模化智能体编排方法论的团队。README 明确这是“builder platform”,需要用户自行设置智能体、定义角色并通过编排器管理团队。若只需单次对话或简单脚本调用,本项目的项目管理层可能偏重。

安装与接入

README 给出的前置条件为 Node.js 18+ 与 Claude CLI(README 链接至 Anthropic 的 Claude Code 文档),并给出安装 Claude CLI 的命令:

npm install -g @anthropic-ai/claude-code

获取与安装源码:

git clone https://github.com/sagiyaacoby/TeamHero.git my-team
cd my-team
npm install

启动方式按平台区分,Windows 执行 launch.bat,Mac/Linux 执行 bash launch.sh。启动后仪表盘地址为 http://localhost:3777,从 Command Center 开始搭建团队。

仓库中存在 .mcp.json,其中配置了 Playwright(npx 启动)与 Figma(HTTP 地址 https://mcp.figma.com/mcp)两个 MCP 服务器;这些属于示例或可选接入配置,与 TeamHero 主体功能的配合方式 README 未展开,需以仓库实际代码核验。项目未提供云端托管入口或在线试用地址,需自行部署。

使用条件与限制

是否需要部署
需自行准备运行环境,具体要求见正文
是否需要注册
账号和认证要求见正文;未明确说明的项目仍待核验
是否需要 API Key
待核验
费用与额度
费用边界见正文,项目许可不代表模型或托管服务免费
所需权限
待核验
许可证与服务条款
待核验

权限费用与限制

  • 项目为自托管,需自行准备 Node.js 环境与 Claude CLI;智能体运行依赖 Claude CLI,其对模型与账号的具体要求以 Anthropic 官方文档为准。
  • 调用 Claude 相关服务可能产生模型使用费用,README 未说明费用结构,需向 Anthropic 核验。
  • 商业产品 Kapow 为独立的托管平台,其定价与服务条款不在本仓库范围内,需向官方核验。
  • 平台会读写项目文件并有文件范围声明机制,实际权限边界与安全影响需自行评估。
  • README 未提供并发规模、系统资源需求或性能数据,相关指标待核验。

国内使用情况

网友核验 用户经验分享

与站长实测独立,不代表官方认证或本站已实测。

我要核验
0 位参与者0 份有效分项记录

最近核验:暂无有效核验。参与人数按当前项目有效身份去重;不同核验项分别统计,不合并成用户总结果。仅统计审核通过且未过期的记录。

分项统计、有效期与样本量
核验项目有效人数成功 / 部分 / 失败 / 无法判断成功率最近核验 / 有效期
官方网站访问00 / 0 / 0 / 0样本不足暂无 / 7 天
GitHub 仓库访问00 / 0 / 0 / 0样本不足暂无 / 7 天
安装与运行00 / 0 / 0 / 0样本不足暂无 / 30 天
官方服务国内连接 / MCP 连接00 / 0 / 0 / 0样本不足暂无 / 30 天
国内网络情况00 / 0 / 0 / 0样本不足暂无 / 7 天
国内模型兼容性00 / 0 / 0 / 0样本不足暂无 / 30 天
自定义 API00 / 0 / 0 / 0样本不足暂无 / 30 天
海外服务依赖
00 / 0 / 0 / 0样本不足暂无 / 30 天
海外账号要求
00 / 0 / 0 / 0样本不足暂无 / 30 天
中文界面
00 / 0 / 0 / 0样本不足暂无 / 30 天
客户端兼容性00 / 0 / 0 / 0样本不足暂无 / 30 天
账号及费用信息
00 / 0 / 0 / 0样本不足暂无 / 30 天

成功率 = 成功 ÷(成功 + 部分成功 + 失败);无法判断不计入分母。至少 5 份可判断结果才显示成功率。账号、费用、海外依赖与语言等事实项,成功表示核查结论与资料吻合,具体观察另行统计;不表示免费或没有海外依赖。不同网络、客户端和版本的结果不能直接等同。

最近有效核验

暂无审核通过且未过期的网友核验。欢迎提交真实使用记录。

查看全部有效核验 →

国内实测状态:国内可用性待核验。协议兼容和官方资料说明不等于国内网络可用。

官方网站访问
尚未实测 待核验
GitHub 仓库访问
尚未实测 待核验
官方服务国内连接
尚未实测 待核验
国内网络情况
尚未实测 待核验
国内模型兼容性
尚未实测 待核验
自定义 API
尚未实测 待核验
海外服务依赖
尚未实测 待核验
海外账号要求
尚未实测 待核验
中文界面
尚未实测 待核验
国内最后核验时间
尚未实测

许可与国内使用

仓库代码以 MIT 许可证发布(LICENSE 文件为 MIT License,Copyright 2024-2026 Sagi Yaacoby)。“TeamHero”与“Kapow”为 Sagi Yaacoby 的商标,详见 TRADEMARK.md。

需区分:MIT 许可仅覆盖本仓库代码,不覆盖 Claude 模型服务、Figma 等第三方 MCP 服务或其托管能力;开源不等于相关云服务免费。README 未说明服务部署地域、国内网络可达性或国内模型支持情况,国内使用可行性与合规性待核验。仓库当前未标记为归档,但项目活跃度与维护计划需自行核实。

项目资料与来源

GitHub 仓库 ↗

官方文档 ↗

字段依据与待核验事项

access:来源确认 · git clone https://github.com/sagiyaacoby/TeamHero.git my-team

project_type:来源确认 · Manage AI agent teams with real project management discipline.

原始 README(折叠技术参考)
# TeamHero

[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
[![Node.js](https://img.shields.io/badge/Node.js-18%2B-339933)](https://nodejs.org/)
[![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](CONTRIBUTING.md)

**Manage AI agent teams with real project management discipline.**

---

## Why This Exists

We have decades of tools for managing people - project boards, task trackers, review workflows. But when it comes to AI agents, there is nothing like that. You spin up agents and hope for the best. No plans, no accountability, no structure. I hit this wall while working on my own projects and could not find anything that treated agent management the way we treat people management.

So I built it. The result became [Kapow](https://gokapow.com) - a managed AI agent platform for businesses. The feedback was immediate. People got it. The idea that agents should work in a structured team with real project management discipline - not just run loose - resonated with developers and business owners alike.

That response made something clear: the need for structured agent orchestration goes way beyond one product. As an entrepreneur, I decided the best thing I could do was release the core architecture as open source to share the unique methodology of running agents at scale. TeamHero is that release - the full orchestration platform under MIT license, no restrictions, nothing held back. My contribution to the open source community.

---

## What It Does

- **Dashboard** - Web UI to manage your entire agent team at a glance
- **Command Center** - Talk to your orchestrator agent through an integrated terminal
- **Task System** - Full lifecycle with plans, versions, approvals, and deliverable tracking
- **Agent Memory** - Every agent has persistent short and long-term memory across sessions
- **Knowledge Base** - Promote deliverables into a searchable, categorized library
- **Skills** - Optional integrations like browser automation and GitHub
- **Conflict Prevention** - File-scope declarations so agents never overwrite each other

---

## What Makes It Different

Most agent frameworks focus on creating agents. TeamHero focuses on managing them. The difference matters when you have multiple agents working on the same project - you need plans before execution, tracked deliverables, conflict prevention, and a single source of truth for what is happening across your team.

TeamHero brings real project management to AI agents. Every task goes through plan, review, execute, and deliver. Every agent has persistent memory. Every file touched is declared upfront so agents do not overwrite each other. It is the discipline that makes agents actually useful at scale - not throwing more of them at the problem.

---

## Who This Is For

Developers who work with Claude Code and want structure around their agents. If you believe in plans before execution, tracked deliverables, and a disciplined approach over "let it run and hope" - this is for you.

This is a builder platform. You set up your agents, define their roles, and manage your team through the orchestrator. It requires comfort with CLI and Node.js.

---

## Quick Start

### Prerequisites

- [Node.js](https://nodejs.org/) 18+
- [Claude CLI](https://docs.anthropic.com/en/docs/claude-code) - the AI backbone that powers your agents

```bash
npm install -g @anthropic-ai/claude-code
```

### Install

```bash
git clone https://github.com/sagiyaacoby/TeamHero.git my-team
cd my-team
npm install
```

### Launch

**Windows:**
```bash
launch.bat
```

**Mac / Linux:**
```bash
bash launch.sh
```

The dashboard opens at `http://localhost:3777`. From there, open the Command Center and start building your team.

---

## Contributing

Contributions are welcome. See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.

---

## License

TeamHero is open source under the [MIT License](LICENSE).

The names "TeamHero" and "Kapow" are trademarks of Sagi Yaacoby. See [TRADEMARK.md](TRADEMARK.md) for details.
历史前置条件说明(补充参考)

Node.js 18+;Claude CLI(README 要求,需另行安装);自托管部署环境;调用 Claude 服务的账号与可能产生的模型费用(待向 Anthropic 核验);可选 MCP 集成如 Playwright、Figma(配置存在,配合方式待核验);国内网络与合规适配情况待核验。

项目来源与清单版权许可

https://github.com/sagiyaacoby/TeamHero

整理来源:https://github.com/slavakurilyak/awesome-ai-agents

清单版权与许可:

MIT License

Copyright (c) 2024 Slava Kurilyak

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.