mcp-agentic-rag Documentation

Technical Documentation: patchy631/mcp-agentic-rag

ℹ️ Provenance: Multi-Tier Ingestion · CodeWiki Reference

Overview (`data/README.md`) Copy Section

Sample Data Directory

This directory is where you should place your documents for the RAG system to ingest.

Supported File Types Copy Chapter

The RAG system uses LlamaIndex's SimpleDirectoryReader, which supports various file types including:

• Text files (.txt)
• PDF files (.pdf)
• Markdown files (.md)
• HTML files (.html)
• JSON files (.json)
• CSV files (.csv)
• And more

Example Usage Copy Chapter

1. Place your documents in this directory
2. Run the server or RAG workflow
3. The documents will be automatically ingested and made available for querying

Sample Document Copy Chapter

For testing purposes, you can create a simple text file in this directory:

# DeepSeek AI

DeepSeek is an AI research company focused on developing large language models.

## DeepSeekR1

DeepSeekR1 is a large language model trained on a diverse corpus of text data. 
It was trained using a combination of supervised learning and reinforcement learning from human feedback (RLHF).

The model architecture is based on a transformer decoder with several billion parameters.
Training was conducted on a large cluster of GPUs over several months.

Key features of DeepSeekR1 include:
- Strong reasoning capabilities
- Code generation and understanding
- Multi-turn conversation abilities
- Knowledge cutoff date of 2023

Save this as deepseek_info.txt in the data directory to test the RAG system with the example query in the main script.

--- METRICS ---

• Files Extracted: 1
• Estimated Token Budget: ~363 tokens
• Canonical Reference: https://codewiki.google/github.com/patchy631/mcp-agentic-rag