Using RAG in Your MCP Client
In this tutorial, you'll learn how to leverage your RAG knowledge base directly from your MCP client (Claude, Cursor, etc.) to enhance your AI assistant with your own documentation and knowledge.
- Plugged.in account with MCP proxy configured
- At least one document uploaded to your Library
- MCP client (Claude Desktop, Cursor, etc.) connected
- Basic understanding of RAG from the 'Building a RAG Knowledge Base' tutorial
1Understanding the pluggedin_rag_query Tool
The pluggedin_rag_query tool is automatically available in your MCP client when connected through Plugged.in. It allows you to search through all documents in your current project's library.
Screenshot: Claude Desktop showing pluggedin_rag_query in available tools
The tool appears in your MCP client's tool list
2Basic RAG Queries
Start with simple queries to search your knowledge base. The tool will return the most relevant chunks from your documents.
Simple Keyword Search
Use the pluggedin_rag_query tool to search for "API authentication"
Screenshot: Claude using RAG query tool
Shows the query being executed and results returned
Question-Based Query
Query my documents about "docker deployment configuration"
Topic Search
Find all mentions of "security best practices" in my uploaded documents
3Advanced Query Techniques
Learn how to craft more sophisticated queries to get better results from your knowledge base.
Based on our project documentation, what are the recommended database indexes for the user table?
The AI will search for database indexing best practices in your documentation
Compare the authentication methods described in our API docs versus the security guidelines document
Great for finding information across multiple documents
Screenshot: Complex RAG query with detailed results
Shows how the AI processes and returns relevant information
4Best Practices
More specific queries yield better results
Include context about what you're trying to achieve for better results
Use RAG queries alongside other MCP tools for powerful workflows:
1. Use pluggedin_rag_query to find deployment steps
2. Follow the instructions found
3. Use pluggedin_send_notification to alert team when complete
Common Issues & Solutions
- Check if documents are uploaded and processed
- Try different keywords or rephrase your query
- Ensure you're in the correct project/profile
- Restart your MCP client
- Check if Plugged.in proxy is running correctly