Getting Started with Plugged.in

Get up and running in 5 minutes. Learn how to set up MCP servers, configure clients, and start building AI-powered applications.

Prerequisites

Node.js v18 or higher (v20+ recommended)
A GitHub or email account for authentication
An MCP-compatible client (Claude Desktop, Cursor, Cline, etc.)
Basic familiarity with command-line interfaces

Quick Start (5 Minutes)

1
Create Your Account

Visit plugged.in and sign up using GitHub (recommended for developers) or email with password.

Plugged.in signup page showing GitHub and email options
Get Started Free
2
Generate Your API Key

After logging in, navigate to Settings → API Keys. Click 'Generate New API Key' and give it a descriptive name.

API key generation interface showing where to create and copy your API key

⚠️ Copy the generated key immediately - you won't see it again!

3
Install the MCP Proxy

Open your terminal and run the following command to install the Plugged.in MCP Proxy:

npx @pluggedin/mcp-proxy@latest --help

This will show you the available options. The proxy acts as a bridge between your MCP client and all your configured servers.

4
Configure Your MCP Client

Add the Plugged.in proxy to your MCP client configuration. Choose your client below:

Configuration file location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)

{
  "mcpServers": {
    "pluggedin": {
      "command": "npx",
      "args": ["-y", "@pluggedin/mcp-proxy@latest"],
      "env": {
        "PLUGGEDIN_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}
5
Test Your Connection

Restart your MCP client and look for the pluggedin tools to confirm the connection is working:

  • pluggedin_discover_tools - Discover capabilities from your servers
  • pluggedin_rag_query - Query your document library
  • pluggedin_send_notification - Send custom notifications
MCP client showing available plugged.in tools

Core Concepts

MCP Servers
MCP servers extend AI capabilities by providing tools (functions the AI can call), resources (data the AI can access), prompts (conversation starters), and custom instructions (server-specific guidance).
Projects & Profiles
Projects are top-level containers for your work. Profiles are workspaces within projects - think of them as environments (dev, staging, production) with isolated server configurations.
Collections
Collections are curated groups of MCP servers that work well together. They're version controlled, shareable with the community, and track installation statistics.
The MCP Proxy
Your single connection point that routes requests to appropriate servers, handles authentication, provides unified logging, and enables advanced features like RAG and notifications.

Your First Steps

1. Add Your First MCP Server

Navigate to 'MCP Servers' in the sidebar and click 'Add New Server'. Choose a server type (NPX, Docker, Binary, or Other).

MCP Server configuration form showing how to add a new server

Example - Add the filesystem server:

Name: Filesystem Access
Type: NPX
Command: @modelcontextprotocol/server-filesystem
Arguments: /path/to/allowed/directory
2. Test in the Playground

The MCP Playground lets you test servers with any AI model. Go to 'Playground' in the sidebar, select your profile and AI model, then start chatting!

MCP Playground interface showing chat interface and server tools
  • Multi-model support via Langchain
  • Virtual scrolling for long conversations
  • Real-time operation logging
  • Custom model instructions
3. Create a Collection

Organize related servers into collections. Navigate to 'Collections', click 'Create New Collection', add servers from your profile, and set visibility preferences.

4. Share with the Community

Make your servers discoverable by clicking 'Share Server' on any configured server. Choose visibility settings, add helpful notes and examples, then track installations and ratings.

Advanced Features

Document Library & RAG
Upload documents to enhance AI context. Navigate to 'Library', upload PDFs, text files, or code. Documents are automatically indexed and queryable via pluggedin_rag_query.
Real-Time Notifications
Stay informed about MCP activities. Click the bell icon to configure preferences, enable email delivery for important alerts, and track all operations.
Custom Instructions
Add server-specific guidance by going to any MCP server's detail page. These instructions guide AI behavior and become available as MCP prompts.
Security Features
Enterprise-grade security with end-to-end encryption (AES-256-GCM), automatic Firejail sandboxing on Linux, API rate limiting, and complete project data isolation.

Troubleshooting

"Session not found" error

  • Ensure your API key is valid
  • Check if the MCP proxy is running
  • Verify your client configuration

Servers not initializing

  • Check server logs in the Playground
  • Verify command paths and arguments
  • Ensure required dependencies are installed

RAG queries not working

  • Confirm documents are uploaded and processed
  • Check if RAG is enabled in Playground settings
  • Verify your subscription includes RAG features

Next Steps

Explore More Guides
API Documentation
Join the Community

Quick Reference

Essential Commands

# Install MCP Proxy npx @pluggedin/mcp-proxy@latest  # Generate encryption key (self-hosted) pnpm generate-encryption-key  # Run migrations (self-hosted) pnpm db:migrate

Configuration Examples

{
  "mcpServers": {
    "pluggedin": {
      "command": "npx",
      "args": ["-y", "@pluggedin/mcp-proxy@latest"],
      "env": {
        "PLUGGEDIN_API_KEY": "your-api-key"
      }
    }
  }
}

Welcome to the Plugged.in community! 🚀

Get Started NowBrowse Documentation