VeriTeknik/pluggedin-mcp-proxy logo

VeriTeknik/pluggedin-mcp-proxy

Free

将多个 MCP 服务器组合到单个接口的综合代理服务器,具有广泛可见性功能 (工具/提示/资源/模板发现管理,调试平台)。

FreeFree tier
Type
Open Source
Company
VeriTeknik

About VeriTeknik/pluggedin-mcp-proxy

Plugged.in MCP Proxy Server is an open-source middleware that aggregates multiple Model Context Protocol (MCP) servers into a single unified interface. It provides a hub with four pillars: Knowledge (RAG v2 for document-aware retrieval with attribution), Memory (persistent, workspace-scoped memory across sessions), Tools (aggregation of built-in and downstream MCP servers' tools, resources, and prompts), and Proxy (single connection for any MCP client). It fetches configuration from the plugged.in App and intelligently routes requests, supporting STDIO (default) and Streamable HTTP with optional API authentication and stateless mode. Compatible with clients like Claude Desktop, Cline, Cursor, and MCP Inspector, it enables centralized management, caching, and telemetry while reducing the need for multiple client configurations.

Key Features

Aggregates multiple MCP servers into a single unified interface
Knowledge feature with RAG v2 for document-aware, attribution-retrieval workflows
Persistent AI memory scoped to workspaces/profiles, surviving sessions
Unified tool discovery with caching and on-demand refresh
Supports STDIO and Streamable HTTP transports
Optional API authentication and stateless mode
Centralized management, policy enforcement, and telemetry
Compatible with Claude Desktop, Cline, Cursor, MCP Inspector, and more
Open source and free to use

Pros & Cons

Pros
  • Reduces the need for multiple MCP client configurations
  • Adds knowledge (RAG) and persistent memory capabilities beyond basic proxying
  • Works with any MCP client without requiring changes to existing setups
  • Open source with an active GitHub repository and documentation
  • Supports both stdio and streamable HTTP for flexible deployment
Cons
  • Requires the plugged.in App for configuration and management
  • Memory tools are still on the roadmap and not yet fully exposed
  • Adds an additional middleware layer that may introduce latency or complexity

Best For

Integrating multiple MCP servers with a single client connectionBuilding document-aware, memory-augmented AI workflowsCentralizing management and monitoring of MCP server configurationsTesting and debugging MCP servers via a unified proxyEnabling cross-session persistence for AI agents and tools

FAQ

What is plugged.in MCP Proxy?
It is an open-source middleware that aggregates multiple Model Context Protocol (MCP) servers into a single unified interface, providing Knowledge, Memory, Tools, and Proxy capabilities.
Which clients are supported?
It works with any MCP client, including Claude Desktop, Cline, Cursor, and MCP Inspector.
Is it free to use?
Yes, the project is open source and free under the provided license.
What are the main pillars of the hub?
The four pillars are Knowledge (document retrieval with attribution), Memory (persistent AI memory), Tools (aggregation of tools/prompts/resources), and Proxy (single connection for all clients).
Does it support persistent memory?
Yes, it provides workspace/profile-scoped persistent memory that survives sessions, with built-in memory tools on the roadmap.