1MCP - One MCP Server for All

Name: 1MCP - One MCP Server for All
Author: 1mcp-app

@1mcp-app

代码仓库主页

关于 1MCP - One MCP Server for All

A unified Model Context Protocol server implementation that aggregates multiple MCP servers into one.

#mcp #mcp-server #unified-mcp-server

基本信息

分类

AI 与智能体

许可证

Apache-2.0

运行时

node

传输方式

stdio

发布者

1mcp-app

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "1mcp": {
      "type": "http",
      "url": "http://127.0.0.1:3050/mcp?app=cursor"
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is 1MCP - One MCP Server for All?

1MCP is a unified MCP runtime that aggregates multiple MCP servers into one process. It addresses configuration sprawl and agent sprawl by providing a single endpoint for clients and a CLI mode for progressive tool discovery, designed for users of Codex, Claude, Cursor, and similar tool-using agents.

How to use 1MCP - One MCP Server for All?

Install globally with npm install -g @1mcp/agent, add upstream servers using 1mcp mcp add, then start the runtime with 1mcp serve. For agent-style workflows, run 1mcp cli-setup --codex (or --claude) to connect the agent to CLI mode, then use 1mcp instructions, 1mcp inspect <server>, and 1mcp run <server>/<tool> --args '<json>' for progressive discovery and execution. Alternatively, use 1mcp proxy for stdio compatibility or direct streamable HTTP for native MCP clients.

Key features of 1MCP - One MCP Server for All

Unified runtime for many MCP servers behind one serve process
CLI mode with progressive discovery: instructions, inspect, run
Template servers for per-client or per-session resolution
Async and lazy loading for faster startup and narrower exposure
Instruction aggregation across static and template servers
Presets, filters, and preset change notifications

Use cases of 1MCP - One MCP Server for All

Give a coding agent one stable runtime with a smaller working surface
Share the same MCP inventory across Cursor, Claude Code, Codex, and internal tooling
Expose context-specific template servers per repo, branch, or session
Centralize auth, filtering, presets, and runtime lifecycle instead of ad hoc scripts

FAQ from 1MCP - One MCP Server for All

What is CLI mode and when should I use it?

CLI mode is the primary workflow for agent-style sessions. It keeps MCP as the backend protocol but narrows what the agent sees at each step using instructions, inspect, and run commands, giving agent loops a smaller working surface.

How does stdio proxy compare to CLI mode?

1mcp proxy provides maximum client compatibility without giving up project context, using the stdio transport that most AI clients already support. It is the recommended fallback after CLI mode, while direct stdio mode is mainly for debugging.

What are template servers?

Template servers are MCP servers materialized from per-client or per-session context, as opposed to static servers loaded at startup. They allow context-specific resolution per repo, branch, or session.

What runtime dependencies are required?

1MCP requires Node.js. Install the package globally via npm: npm install -g @1mcp/agent.

Can I connect a native MCP HTTP client directly?

Yes. Direct streamable HTTP MCP access is supported for clients that already speak MCP natively and do not need project context, using a URL like http://127.0.0.1:3050/mcp?app=cursor.

AI 与智能体分类下的更多 MCP 服务器

Solon Ai

opensolon

Java AI application development framework (supports LLM-tool,skill; RAG; MCP; Agent-ReAct,Team-Agent). Compatible with java8 ~ java25. It can also be embedded in SpringBoot, jFinal, Vert.x, Quarkus, and other frameworks.

416