Mcproxy F99b2898

MCP开发工具

提供方 LaurentAjdnik

开发工具 stdio · 本地进程 无需鉴权 社区 / 第三方 Claude

A proxy between MCP Clients and MCP Servers

一句话选型结论

按资料字段整理

Mcproxy F99b2898 属于开发工具类 MCP 服务器,stdio · 本地进程,无需鉴权,可直接接入 Claude,没有现成安装命令,需按官方文档接入。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。

分类开发工具 传输方式stdio · 本地进程 鉴权无需鉴权 部署方式本地部署
GitHub 星标 2 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

开发工具
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

developer-tools代理服务本地部署模型协议插件扩展日志记录能力聚合

安装与配置

stdio · 本地进程 无需鉴权
客户端在本机拉起子进程,数据不出本机
  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 Node.js 等运行时,再按安装命令拉取依赖。

  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
代码与仓库操作

在对话里直接查代码、提 issue、走 PR,省掉来回切工具。

本地命令与脚本

把构建、测试、格式化这类重复命令交给模型执行。

接口联调与测试

让模型调你的接口、跑一遍用例,快速定位问题。

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

  • 已适配 Claude,换客户端不用重新研究配置。
  • 资料标注无需鉴权,装好就能连。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 没有现成安装命令,需要按官方文档自己填启动命令和参数。
  • 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

  • 需要在开发工具场景里让模型直接调用外部能力的人。
  • 已经在用 Claude 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

  • 完全不想在本地装环境的人。
  • 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
Mcproxy F99b2898当前 未标注 stdio · 本地进程 本地部署 无 2026-05-18
Git 838e2ec1 12 stdio · 本地进程 本地部署 有 2026-05-17
Markitdown MCP 2c653a5f 1 stdio · 本地进程 本地部署 有 2026-05-17
JavaGuide 未标注 stdio · 本地进程 本地部署 有 2026-04-11
awesome-claude-skills 未标注 stdio · 本地进程 本地部署 有 2026-04-26
Context7 831acf40 未标注 stdio · 本地进程 本地部署 有 2026-05-17

服务器档案

开发工具
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LaurentAjdnik上游标注的维护方 LaurentAjdnik
分类开发工具本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式无需鉴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 · 未标注本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2026-05-18上游最后核验时间

详细介绍

6.8 KB
来自项目仓库与上游目录的原文,未做删改

Foreword

This is very experimental and possibly totally useless 😁. Join this discussion for feedback, that will be greatly appreciated.

Note: This might contradict this: "Cross-server interactions are controlled by the host". Or not...

Overview

This project is related to Anthropic's Model Context Protocol (MCP, see docs and spec).

MCProxy is a proxy between MCP Clients and MCP Servers, allowing new features in the workflow between them (a few examples are given below).

MCProxy provides the building blocks for such a proxy architecture, along with some basic features (TBD: List of basic features).

New features can be added, depending on the user's needs, with plug-in modules (TBD: Plug-in mechanism).

Global architecture

From an MCP Client's perspective, MCProxy behaves like an MCP Server. From an MCP Server's perspective, MCProxy behaves like an MCP Client.

In its simplest form:```mermaid
flowchart LR
client1[MCP Client]
subgraph "MCProxy"
server1[MCP Server]
client2[MCP Client]
end
server2[MCP Server]

client1 <--MCP--> server1
client2 <--MCP--> server2
server1 <--> client2

An MCProxy can connect to multiple servers:```mermaid
flowchart LR
    client1[MCP Client]
    subgraph "MCProxy"
        server1[MCP Server]
        client2[MCP Client A]
        client3[MCP Client B]
    end
    server2[MCP Server A]
    server3[MCP Server B]

    client1 <--MCP--> server1
    client2 <--MCP--> server2
    client3 <--MCP--> server3
    server1 <--> client2
    server1 <--> client3

An MCP Host may connect to different MCProxies with very different configurations:```mermaid
flowchart LR
subgraph "Host"
clientH1[MCP Client]
clientH2[MCP Client]
end
subgraph "MCProxy A"
serverPA[MCP Server]
clientPA[MCP Client]
end
serverA[MCP Server A]
subgraph "MCProxy B"
serverPB[MCP Server]
clientPB[MCP Client]
end
serverB[MCP Server B]

clientH1 <--MCP--> serverPA
clientH2 <--MCP--> serverPB
clientPA <--MCP--> serverA
clientPB <--MCP--> serverB
serverPA <--> clientPA
serverPB <--> clientPB

If of any use, MCProxies can be chained:```mermaid
flowchart LR
    client1[MCP Client]
    subgraph "MCProxy A"
        server1[MCP Server]
        client2[MCP Client]
    end
    subgraph "MCProxy B"
        server2[MCP Server]
        client3[MCP Client]
    end
    server3[MCP Server]

    client1 <--MCP--> server1
    client2 <--MCP--> server2
    client3 <--MCP--> server3
    server1 <--> client2
    server2 <--> client3

Internal architecture

The basic components are:
- The Server manager, that handles the communication with the MCP Client (Requests but also Responses/Errors/Notifications)
- The Clients manager, that maintains a list of connected MCP Servers and handles the communication with them (Requests but also Responses/Errors/Notifications)

The Internal features component provides a list of basic features that are always available.

The Modules manager maintains a list of external modules, which provide extra-features, and interacts with them.

The core component is the Dispatcher, that:
- Forwards messages between the MCP Client and the MCP Servers, back and forth
- Applies the internal features, when applicable
- Calls the external modules, when applicable```mermaid
flowchart LR
subgraph MCProxy
subgraph servermanager[Server manager]
server[MCP Server]
end
dispatcher[Dispatcher]
internalfeatures[Internal features]
modulesmanager[Modules manager]
subgraph clientsmanager[Clients manager]
direction TB
clientA[MCP Client A]
clientB[MCP Client B]
end
end
moduleA[External module A]
moduleB[External module B]

servermanager <--> dispatcher
dispatcher <--> clientsmanager
dispatcher <--> internalfeatures
dispatcher <--> modulesmanager
modulesmanager <--> moduleA
modulesmanager <--> moduleB

```

Identifiers

By default, MCProxy exposes all the resources/prompts/tools of the MCP Servers it connects to, using the same identifiers (names, URIs...). These identifiers can be renamed (see Features below).

Configuration

In order to connect to MCP Servers, MCProxy uses the same configuration file format as defined in the Quickstart section of the MCP docs.

Some more internal configuration is available for the MCProxy itself.

Finally, each module provides some configuration directives. Examples are given below.

Note: In MCP Hosts, like Claude Dekstop, the configuration file would point towards the MCProxy, not the different servers behind it.

Features

Here are a few examples of features an MCProxy could implement, internally of through external modules.

Logging

An MCProxy can log all messages going back and forth between an MCP Client and an MCP Server.

Configuration:
- What to log
- Where to log

Capabilities aggregation

An MCProxy can aggregate capabilities from different MCP Servers into a meaningful package.

Capabilities blocking

An MCProxy may expose only a subset of the capabilities provided by the MCP Server(s) it connects to.

Configuration:
- Capabilities to be blocked

Capabilities disambiguation

With the fast-growing number of MCP Servers, me might end up with some name collisions. An MCProxy may rename capabilities, so that they appear unique to an MCP Client, and handle routing to the appropriate MCP Server.

Configuration:
- List of Server / Identifier => New identifier

Content update

Textual content can be filtered or enriched while going through MCProxy.

On-the-fly internationalization (i18n)

The current version of the MCP Specification does not provide i18n (see my issue about this).

Configuration:
- Preferred language

External modules

They provide extra-features on top of the basic features provided by MCProxy itself.

They are independent projects that plug into an MCProxy.

A list of such modules could be maintained, just like lists of MCP Clients/Servers are currently being built (see for instance here, here or here).

Hosting

They can be hosted locally or remotely as a service.

Capabilities

They declare their "capabilities", that it the types of MCP messages they are interested in.

They are called in the order in which they are declared in the configuration file.

Protocol

The protocol between MCProxy and an external module is strongly inspired by the MCP specification.

But the workflow is quite different: Mostly, a response usually has the same type as the message sent.

For instance:
- a GetPromptRequest message sent to a module would return an updated GetPromptRequest message
- a ListToolsResult message sent to a module would return an updated ListToolsResult message

(TBD: Specification of such a protocol)

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
Mcproxy F99b2898 怎么安装?

资料没有给出安装命令,请到官方仓库或文档按说明接入;这类条目多数是生态应用,或尚未发布可安装包。

Mcproxy F99b2898 怎么在 Claude 里配置?

它的传输方式是stdio · 本地进程,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。

Mcproxy F99b2898 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。

Mcproxy F99b2898 需要 API Key 吗?

资料标注为无需认证,装好就能连;如果实际调用时报鉴权错误,按官方文档补上凭据。

Mcproxy F99b2898 支持哪些客户端?

资料标注支持 Claude;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。

Mcproxy F99b2898 能提供哪些能力?

资料显示能力标签包括 代理服务、Claude、日志记录。具体工具清单以官方文档为准。

Mcproxy F99b2898 需要什么运行环境?

属于本地进程,需要在本机装好运行环境。

Mcproxy F99b2898 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

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 2026-05-18。