DeployStack

MCP其他

提供方 deploystackio

其他 HTTP · 远程 OAuth 授权 社区 / 第三方 ClaudeCursorVS CodeWindsurf
查看接入方式

Open source MCP hosting - deploy MCP servers to HTTP endpoints for n8n, Dify, Voiceflow, and any MCP client.

一句话选型结论

按资料字段整理

DeployStack 属于其他类 MCP 服务器,HTTP · 远程,OAuth 授权,可直接接入 Claude、Cursor、VS Code 等,有可直接复制的远程端点配置。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类其他 传输方式HTTP · 远程 鉴权OAuth 授权 部署方式远程部署
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

其他
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

nodejsmcpmodel context protocol

安装与配置

HTTP · 远程 OAuth 授权
连远程 HTTPS 端点,通常为 Streamable HTTP
https://satellite.deploystack.io/mcp

把它填进客户端的 MCP 配置即可,不需要在本地安装运行时。

  1. 1
    准备运行环境

    确认你的客户端支持远程端点,并拿到服务地址。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
接入外部数据

把外部系统的数据接进对话。

自动化重复操作

把固定步骤交给模型跑。

补充工具能力

给模型补上它原本没有的能力。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude、Cursor、VS Code,换客户端不用重新研究配置。

先想清楚的地方

  • 远程端点依赖网络与服务方可用性,注意密钥与调用额度。
  • 使用前要准备好OAuth 授权对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在其他场景里让模型直接调用外部能力的人。
  • 已经在用 Claude、Cursor 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
DeployStack当前 未标注 HTTP · 远程 远程部署 有 未标注
gemini-cli 未标注 stdio · 本地进程 本地部署 有 2026-04-13
ext-apps 未标注 stdio · 本地进程 本地部署 有 2026-04-19
northcinder 未标注 stdio · 本地进程 本地部署 有 2026-08-23
fastctx 未标注 stdio · 本地进程 本地部署 有 2026-08-23
quickstart-resources 未标注 stdio · 本地进程 本地部署 有 2026-04-12

服务器档案

其他
字段来自项目仓库与公开索引,中文说明由本站标注
提供方deploystackio上游标注的维护方 deploystackio
分类其他本站按上游分类展示
连接方式HTTP · 远程连远程 HTTPS 端点,通常为 Streamable HTTP
鉴权方式OAuth 授权远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · node本地安装时用到的包标识与运行时
部署类型远程部署上游给出的部署形态
最近核验未标注上游最后核验时间

详细介绍

12.2 KB
来自项目仓库与上游目录的原文,未做删改

DeployStack - Open Source MCP Hosting Platform

🚀 Try it out ·
🌐 Website ·
📚 Documentation ·
 Roadmap  ·
 Discord

Deploy MCP servers from GitHub to HTTP endpoints in 30 seconds. Works with n8n, Dify, Voiceflow, Langflow, Claude Code, Cursor, and any MCP client.

The Problem

Most MCP servers are stdio-only. But workflow automation platforms need HTTP endpoints.

Platform What They Need The Problem
n8n HTTP URLs for MCP Client node stdio doesn't work
Dify HTTP/SSE endpoints (v1.6+) No stdio support
Voiceflow Remote MCP endpoints "Local MCP NOT supported"
Langflow HTTP/SSE for production Local processes won't scale

Current workarounds are painful:

  • mcp-remote requires local Node.js setup
  • Docker containers need DevOps expertise
  • DIY hosting (Fly.io, Railway) takes hours

And for teams, the problems multiply:

  • Credentials scattered across Slack DMs and .env files
  • No visibility into who uses which tools
  • Onboarding new developers takes hours
  • When someone leaves, good luck rotating all those API keys

Who DeployStack Is For

For AI Automation Builders

  • n8n, Dify, Voiceflow, Langflow users: Get HTTP endpoints for MCP servers - no more stdio limitations
  • No DevOps required: Skip the Docker containers and mcp-remote setup
  • Deploy your custom MCP servers: Point at a GitHub repo, get a URL in 30 seconds
  • Focus on workflows: Build automations, not infrastructure

For Automation Agencies

  • Client-ready MCP hosting: Deploy MCP servers for client projects without managing infrastructure
  • White-label ready: Team isolation keeps client data separate
  • Scale without ops burden: Add more MCP servers without adding DevOps headcount
  • Credential management: Secure vault for all your clients' API keys

For Platform / DevOps Teams

  • End credential chaos: Move API keys from Slack DMs and .env files into an encrypted vault with automatic injection
  • Control MCP access: Role-based permissions determine who can use which MCP servers and tools
  • Prevent data exposure: Team isolation and audit logging show exactly what tools accessed what data
  • Centralize configuration: One source of truth instead of N developers × M configurations

For Developers

  • Zero setup: Add one URL to your MCP client, get access to all team MCP tools instantly
  • No local installs: No npm install, no process management, no port conflicts
  • Just works: Same configuration everywhere - no more "works on my machine" debugging
  • Focus on code: Security and credentials are handled for you

For Engineering Managers

  • 1-click team onboarding: New developer? They're productive with MCP tools in minutes, not hours
  • Reduce AI costs: 98% token reduction means lower API bills and better LLM performance
  • Full visibility: See which tools your team uses, how often, and what they access
  • Offboarding safety: When someone leaves, revoke access instantly - no credential rotation nightmare

The Solution

DeployStack hosts MCP servers as HTTP endpoints. Two ways to use it:

Deploy from GitHub

Point DeployStack at any GitHub repo with an MCP server:

  1. Connect your GitHub account
  2. Select a repository
  3. Get an HTTP endpoint that works immediately

Auto-deploys when you push. Like Vercel, but for MCP servers.

Use the Curated Catalog

Don't have your own MCP server? Browse our catalog of popular servers:

  • One-click installation
  • Pre-configured and tested
  • Ready to use in seconds

Key Features

Deploy from GitHub```bash

GitHub Repo → DeployStack → HTTP Endpoint → Your MCP Client


- Any stdio MCP server becomes an HTTP endpoint
- Auto-deploy on git push
- Branch deployments for testing
- Zero DevOps required

### 98% Token Reduction

Traditional MCP setup loads all tools into context:

- 10 servers × 15 tools = 150 tools = 75,000 tokens consumed

DeployStack's hierarchical router exposes only 2 meta-tools:

- `discover_mcp_tools` - find relevant tools
- `execute_mcp_tool` - run the tool you need
- Result: 1,372 tokens instead of 75,000

Scale to 100+ MCP servers without degrading LLM performance.

### Credential Vault

Store API keys and secrets securely:

- Encrypted at rest
- Automatic injection at runtime
- No credentials in config files
- No secrets in Slack DMs
- Rotate once, update everywhere

### Team Management & RBAC

Control access across your team:

- Invite team members with specific roles
- Define who can use which MCP servers
- Manage permissions from one dashboard
- Instant onboarding for new developers

### Audit Logging

Full visibility into MCP usage:

- Track every tool interaction
- See who accessed what and when
- Export logs for compliance
- Debug issues with complete history

### Works Everywhere

Compatible with any MCP client that supports HTTP/SSE:

| Workflow Platforms | AI Coding Tools | Other Clients       |
| ------------------ | --------------- | ------------------- |
| n8n                | Claude Code     | Any HTTP MCP client |
| Dify               | Cursor          | Custom integrations |
| Voiceflow          | VS Code         |                     |
| Langflow           | Windsurf        |                     |

## Architecture```text
┌─────────────┐     ┌─────────────────┐     ┌─────────────┐
│   GitHub    │────▶│   DeployStack   │────▶│  HTTP URL   │
│    Repo     │     │    Satellite    │     │   Endpoint  │
└─────────────┘     └─────────────────┘     └─────────────┘
                            │
                    ┌───────┴───────┐
                    │               │
              ┌─────▼───-──┐   ┌───-──▼──-───┐
              │ Credential │   │ Hierarchical│
              │   Vault    │   │   Router    │
              └──────────-─┘   └─--──────────┘

Two deployment options:

Global Satellites Team Satellites
Managed by DeployStack Deploy on your infrastructure
Zero setup Full data control
Free tier available Enterprise security
satellite.deploystack.io satellite.yourcompany.com

Quick Start

1. Sign Up

Create a free account at cloud.deploystack.io

2. Add MCP Servers

Option A: Deploy from GitHub

  • Connect your GitHub account
  • Select a repository with an MCP server
  • Click Deploy

Option B: Use Catalog

  • Browse the MCP server catalog
  • Click Install on any server
  • Configure credentials if needed

3. Get Your Endpoint

Copy the HTTP endpoint URL from your dashboard.

4. Configure Your MCP Client

Add to your MCP client configuration:```json
{
"mcpServers": {
"deploystack": {
"type": "http",
"url": "https://satellite.deploystack.io/mcp"
}
}
}


That's it. All your MCP servers are now accessible via one URL.

## Self-Hosting

DeployStack is open source (AGPL-3.0). Run it on your own infrastructure:```bash
git clone https://github.com/deploystackio/deploystack
cd deploystack
docker-compose up

See Self-Hosting Documentation for detailed instructions.

Roadmap

Phase 1: Foundation (Completed)

  • [x] Deployed cloud.deploystack.io hosted version with backend and frontend
  • [x] Implemented secure user and team management system with roles and permissions
  • [x] Integrated OAuth for secure logins (GitHub)
  • [x] Created initial MCP Server Catalog for tool discovery
  • [x] Established documentation and self-hosted Docker support
  • [x] PostgreSQL-only database architecture for enterprise scalability

Phase 2: Enterprise Governance (Completed)

  • [x] Auto-install MCP servers for new users with admin-controlled defaults
  • [x] Featured MCP servers filtering for improved tool discovery
  • [x] Global Event Bus System with plugin integration
  • [x] Three-Tier MCP Configuration Architecture (template, team, user levels)
  • [x] Multi-User Configuration Management with personalized device-specific configs
  • [x] Advanced MCP Argument & Environment Variable Handling with schema validation
  • [x] HTTP Headers & URL Query Parameters Support for HTTP/SSE transport
  • [x] Selective Secret Masking in Satellite Logs

Phase 3: Satellite Infrastructure (Completed)

  • [x] Global Satellite Infrastructure - managed MCP servers via HTTPS
  • [x] Zero-Installation Experience - just add URL to your MCP client
  • [x] OAuth Authentication with token-based auth
  • [x] Satellite Pairing Security with JWT-based token registration
  • [x] Public Launch - production satellite for community use
  • [x] Resource Management - process isolation and limits
  • [x] GitHub README and Stars Integration for MCP catalog
  • [x] Background Job Queue System with PostgreSQL-based worker infrastructure
  • [x] Cron Job Scheduling System with node-cron integration
  • [x] MCP Registry Integration - official MCP Registry support

Phase 4: Context Window Optimization (Completed)

  • [x] Hierarchical Router Implementation - 2-tool pattern achieving 98% token reduction
  • [x] Semantic Tool Search with Fuse.js fuzzy matching (2-5ms performance)
  • [x] Tool Discovery API with paths, descriptions, transport types, relevance scores
  • [x] Smart Tool Execution with automatic routing to stdio or HTTP/SSE backends
  • [x] MCP Tool Metadata Collection with token counting and event-driven sync
  • [x] Tool Display & Analytics API for team-aware tool management
  • [x] Token Analytics Dashboard with visual comparisons and per-installation breakdown
  • [x] OAuth 2.1 Authentication for MCP Servers (Box, Linear, GitHub Copilot, etc.)

Phase 5: Advanced Governance (Completed)

  • [x] Analytics dashboards for tool usage and performance
  • [x] Per-tool access controls - team admins can disable individual tools
  • [x] Improved searchable MCP Server Catalog with Featured page and categories
  • [x] AI Instructions - ready-to-copy instruction files for AI coding assistants

Phase 6: Advanced Architecture (Completed)

  • [x] Multi-Transport Support - SSE, Streamable HTTP, Direct HTTP protocols
  • [x] Real-Time Command Orchestration with instant status feedback
  • [x] Satellite Job System with JobManager and extensible architecture
  • [x] Satellite Backend Events System with batch processing and JSON schema validation
  • [x] MCP Client Activity Tracking across all components
  • [x] Time-Series Metrics System with 15-minute bucket aggregation
  • [x] Multi-region satellite deployment

Phase 7: GitHub Deployment (Completed)

  • [x] Deploy MCP servers from any GitHub repository
  • [x] Auto-deploy on git push via webhooks
  • [x] Branch deployments for testing
  • [x] Team Satellites - customer-deployed satellites for enterprise

Phase 8: In Progress

  • [ ] Advanced Audit UI - detailed logging dashboard
  • [ ] Python MCP Server Support - deploy Python-based MCP servers

Phase 9: Planned

  • [ ] SSO Integration (SAML, OIDC)
  • [ ] SOC 2 Type II certification
  • [ ] AI-powered tool recommendations based on conversation context
  • [ ] Per-team token budgets and cost attribution

Contributing

We welcome contributions. Here's how:

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/your-idea
  3. Follow TypeScript and ESLint conventions
  4. Add tests for new functionality
  5. Submit a pull request

Areas where we need help:

  • MCP server integrations
  • Documentation improvements
  • Satellite infrastructure
  • Enterprise features

Links

  • Platform: cloud.deploystack.io
  • Documentation: docs.deploystack.io
  • GitHub: github.com/deploystackio/deploystack
  • Discord: Join the community

License

AGPL-3.0 - you can self-host, modify, and use DeployStack freely. If you offer it as a service, you must open-source your modifications.

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
DeployStack 怎么安装?

在客户端按下面的安装命令接入:https://satellite.deploystack.io/mcp。装完在客户端配置里引用同一个包或端点。

DeployStack 怎么在 Claude、Cursor、VS Code 里配置?

它的传输方式是HTTP · 远程,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

DeployStack 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

DeployStack 需要 API Key 吗?

资料标注需要OAuth 授权认证,使用前准备好对应凭据,并按最小权限配置。

DeployStack 支持哪些客户端?

资料标注支持 Claude、Cursor、VS Code、Windsurf、Codex、Zed;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

DeployStack 能提供哪些能力?

资料显示能力标签包括 nodejs、mcp、model context protocol、mcp server。具体工具清单以官方文档为准。

DeployStack 需要什么运行环境?

运行时为 node;属于远程端点,不需要在本地装运行时。

DeployStack 是本地还是远程?

它是远程端点,连服务地址即可。传输方式为HTTP · 远程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 未标注。