能力标签

🔌 MCP 🤖 Agent 🔄 工作流 💻 CLI 🔗 REST API 🧠 Claude

🔌

MCP工具

开源MCP工具

基于 TypeScript · 让 AI 助手直接操作你的系统与工具

英文名：cyborg

⭐ 17 Stars 💻 TypeScript 📄 NOASSERTION 🏷 AI 8.0分

8.0AI 综合评分

aiagent-platformmcp

⚙️ 配置说明 📺 TG 频道

✦ AI Skill Hub 推荐

经 AI Skill Hub 精选评估，开源MCP工具获评「强烈推荐」。这款MCP工具在功能完整性、社区活跃度和易用性方面表现出色，AI 评分 8.0 分，适合有一定技术背景的用户使用。

📚 深度解析

开源MCP工具是一款基于 MCP（Model Context Protocol）标准协议的 AI 工具扩展。MCP 协议由 Anthropic 开发并开源，旨在建立 AI 模型与外部工具之间的标准化通信接口，目前已被 Claude Desktop、Claude Code、Cursor 等主流 AI 工具采纳。

通过安装开源MCP工具，你的 AI 助手将获得额外的工具调用能力，可以用自然语言直接操控该工具的功能，无需学习复杂的命令行语法。MCP 工具的核心价值在于"一次配置，永久增强"——配置完成后，每次与 AI 对话时都可以无缝调用这些工具。

在技术实现上，MCP 工具通过标准的 JSON-RPC 协议与 AI 客户端通信，工具的功能以"工具列表"的形式暴露给 AI 模型，AI 可以按需调用。开源MCP工具提供了结构化的工具调用接口，使 AI 模型能够精确地理解和使用每个功能点，显著降低 AI 在工具使用上的错误率。

与传统的 API 集成相比，MCP 工具的优势在于无需编写代码——用户只需在配置文件中添加几行 JSON，即可让 AI 获得全新能力。AI Skill Hub 将开源MCP工具评为 AI 评分 8.0 分，属于同类工具中的优质选择。

📋 工具概览

开源MCP工具是一款遵循 MCP（Model Context Protocol）标准协议的 AI 工具扩展。通过 MCP 协议，它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务，实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用，都可以通过自然语言在 AI 对话中直接触发，极大提升生产效率。

GitHub Stars

⭐ 17

开发语言

TypeScript

支持平台

Windows / macOS / Linux

维护状态

轻量级项目，按需更新

开源协议

NOASSERTION

AI 综合评分

8.0 分

工具类型

MCP工具

Forks

—

📖 中文文档

以下内容由 AI Skill Hub 根据项目信息自动整理，如需查看完整原始文档请访问底部「原始来源」。

📌 核心特色

通过标准 MCP 协议与 Claude、Cursor 等主流 AI 客户端深度集成
提供结构化工具调用接口，显著降低 AI 集成复杂度
支持 Claude Desktop 和 Claude Code 无缝接入，开箱即用
可与其他 MCP 工具组合叠加，构建完整 AI 工作站
轻量无侵入设计，不影响现有系统架构

🎯 主要使用场景

在 Claude Desktop 对话中直接调用本地工具，实现 AI 与系统的深度联动
通过自然语言驱动复杂的多步骤自动化任务，代替繁琐手动操作
将多个 MCP 工具组合使用，构建个人专属 AI 工作站

以下安装命令基于项目开发语言和类型自动生成，实际以官方 README 为准。

安装命令

# 方式一：通过 Claude Code CLI 一键安装
claude skill install https://github.com/Cyborg7-com/cyborg

# 方式二：手动配置 claude_desktop_config.json
{
  "mcpServers": {
    "--mcp--": {
      "command": "npx",
      "args": ["-y", "cyborg"]
    }
  }
}

# 配置文件位置
# macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows: %APPDATA%/Claude/claude_desktop_config.json

📋 安装步骤说明

确认已安装 Node.js（v18 或以上版本）
打开 Claude Desktop 或 Claude Code 的 MCP 配置文件
按「交给 Agent 安装 → Claude Desktop」标签中的 JSON 配置填入 mcpServers 字段
保存配置文件并重启 Claude 客户端
重启后，在对话中即可使用本工具

以下用法示例由 AI Skill Hub 整理，涵盖最常见的使用场景。

常用命令 / 代码示例

# 安装后在 Claude 对话中直接使用
# 示例：
用户: 请帮我用 开源MCP工具 执行以下任务...
Claude: [自动调用 开源MCP工具 MCP 工具处理请求]

# 查看可用工具列表
# 在 Claude 中输入："列出所有可用的 MCP 工具"

以下配置示例基于典型使用场景生成，具体参数请参照官方文档调整。

配置示例

// claude_desktop_config.json 配置示例
{
  "mcpServers": {
    "__mcp__": {
      "command": "npx",
      "args": ["-y", "cyborg"],
      "env": {
        // "API_KEY": "your-api-key-here"
      }
    }
  }
}

// 保存后重启 Claude Desktop 生效

📑 README 深度解析真实文档完整度 61/100 含工作流图查看 GitHub 原文 →

以下内容由系统直接从 GitHub README 解析整理，保留代码块、表格与列表结构。

简介

Cyborg7

Collaborative AI agents — humans and agents on different machines, working together in shared workspaces.

Each person runs their own daemon, agents execute locally, and a relay broker connects everyone. No cloud runs your code.

<a href="#quick-start">Quickstart</a> · <a href="docs/README.md">Docs</a> · <a href="#connect-your-agents">Providers</a> · <a href="#cybos">Cybos</a> · <a href="#architecture">Architecture</a> · <a href="#develop">Develop</a> · <a href="CONTRIBUTING.md">Contributing</a>

---

Cyborg7 is a distributed, multi-daemon platform where humans and AI agents are teammates in the same workspace. Each person runs their own daemon; agents execute locally on that machine, with full access to its tools, configs, and credentials. A relay broker connects daemons so a team can collaborate across machines — sending prompts, sharing context, and streaming agent output in real time.

There is no central server that runs your agents and no cloud execution of your code. Your agents live where your work lives. PostgreSQL holds the shared state every teammate sees — workspaces, channels, messages, tasks — while each daemon keeps a local SQLite cache for fast, offline-friendly reads.

It is built as a fork of Paseo, extending it with workspaces, channels, tasks, and Cybos (custom AI personalities). Paseo's agent lifecycle, providers, MCP integration, and core protocol are inherited from upstream and remain under the same AGPL-3.0 license. See NOTICE for attribution.


🖥️ Distributed local daemons	Every machine runs its own daemon. Agents are spawned as local child processes with Cyborg7 MCP tools injected — full access to your tools and credentials, nothing executes in the cloud.
🔌 Multi-provider	Drive Claude Code, Codex, Copilot, OpenCode, and Pi through one interface. Pick the right model and harness per agent.
💬 Workspaces & channels	Organize work into workspaces and channels. Every message — human or agent — is persisted to shared storage and visible to the whole team.
🧬 Cybos	Custom AI personalities with their own identity, system prompt, and model preferences. Define a `cybo.json` + `soul.md` pair and run it standalone or inside any workspace.
🖧 Cross-device	Desktop (Electron), web (Svelte 5), and a terminal CLI, all driving the same daemon and shared state.
🔒 Self-hosted, no telemetry	Run the whole thing on your own machines. PostgreSQL for shared state, SQLite for local cache. No telemetry, no forced logins.

[!NOTE] Cyborg7 is free and open source under AGPL-3.0. If you run a modified version as a network service, the AGPL requires you to offer that modified source to its users. See LICENSE and NOTICE.

---

What's in the box

Runtime — Node.js 22+, TypeScript (strict)
UI — Svelte 5 (runes) + Tailwind CSS v4 + shadcn-svelte
Transport — Hono HTTP + WebSocket; relay pub/sub between daemons
Shared state — PostgreSQL via Drizzle ORM
Local cache — SQLite via better-sqlite3
Agent SDKs — @anthropic-ai/claude-agent-sdk (Claude) and ACP over stdio (Codex, Qwen, …)
Assets — S3 with presigned-URL uploads
Relay scale-out — Redis (optional) for pub/sub between relay instances
Desktop — Electron shell (connects to a cloud relay)
Tooling — oxlint + oxfmt

Deployments

Cyborg7 has two deployment shapes that run from the same code:

Local daemon — the full daemon with SQLite (and optional PostgreSQL). Agents run locally on your machine.
Cloud relay — relay-standalone.ts is a standalone broker (Hono HTTP + WebSocket) for teams whose daemons need to reach each other across networks. It brokers messages and queries shared PostgreSQL directly; it does not run agents — those always stay on each user's own daemon.

---

Quick start

git clone https://github.com/Cyborg7-com/cyborg.git
cd cyborg
pnpm install
pnpm dev

pnpm dev starts the daemon on port 6780 and the UI dev server on 5173. Open the UI, and you have a Slack-style workspace driven by your local daemon.

Environment

cp packages/server/.env.example packages/server/.env

Variable	What it does
`DATABASE_URL`	PostgreSQL connection string. Setting it switches the daemon into connected (multi-user) mode; leaving it unset keeps it solo on SQLite.
`CYBORG7_JWT_SECRET`	JWT signing secret. Required in production — the daemon refuses to boot with the development default outside development.

Solo vs. connected

Cyborg7 runs in two modes, auto-detected from your environment:

Solo — no DATABASE_URL. The daemon uses SQLite only. Best for working alone or developing against a single machine.
Connected — set DATABASE_URL. The daemon keeps its SQLite cache and also writes through to a shared PostgreSQL, so teammates' daemons share workspaces, channels, messages, and tasks. The relay brokers messages between daemons.

🎯 aiskill88 AI 点评 A 级 2026-06-27

开源MCP工具，具有较高的潜力和应用价值