theaiautomators

Claude Code Agentic Rag Series

Name: Claude Code Agentic Rag Series
Author: theaiautomators

Data & AI community intermediate

SOURCE https://github.com/theaiautomators/claude-code-agentic-rag-series

Description

Planning docs and PRDs for the Claude Code Agentic RAG Masterclass video series. Build a full-stack AI app with Python, React, and Supabase.

Installation

Terminal

claude install-skill https://github.com/theaiautomators/claude-code-agentic-rag-series

README

The AI Automators

Claude Code Agentic RAG Series

A reference implementation and learning foundation for building **agentic RAG systems with AI coding tools**.

This project was built collaboratively with Claude Code across multiple video episodes, where each episode adds new capabilities. This public repository contains the **PRDs, prompts, and planning documents** used to build each episode.

Claude Code RAG Masterclass

The Episodes

Episode	Title	What It Adds
Episode 1	Agentic RAG Masterclass	Full RAG foundation: chat UI, document ingestion, hybrid search, tool calling, sub-agents
Episode 2	Knowledge Base Explorer	Filesystem-like tools for hierarchical knowledge navigation
Episode 3	PII Redaction & Anonymization	Privacy layer ensuring sensitive data never reaches cloud LLMs
Episode 4	Agent Skills & Code Sandbox	Reusable skills system + Python execution in Docker containers
Episode 5	Advanced Tool Calling	Dynamic tool registry, sandbox bridge, MCP integration
Episode 6	Agent Harness & Workflows	Deep Mode autonomous agent, domain-specific harness engine, contract review workflow

The App

The Agentic RAG App is a full-stack AI application built on **Python, React, and Supabase**.

It includes user authentication, a chat interface, document ingestion, and a complete RAG pipeline with hybrid search and reranking. The chat supports agentic features like text-to-SQL, web search, sub-agents, and a knowledge base explorer with filesystem-like tools.

Advanced features include an Agent Skills system, a code sandbox for executing Python in isolated Docker containers, PII redaction for removing sensitive data, and a full autonomous agent platform with Deep Mode for multi-step tasks and a domain-specific harness engine for deterministic workflows (including a contract review harness with DOCX report generation).

Choose Your Path

Path 1: Build from Scratch

The deepest learning experience. You're not starting with any scaffold—you're building the entire system yourself.

**Get started:** Watch the [YouTube episodes](https://www.youtube.com/watch?v=4Tp6nPZa5is&list=PLmc6yUBIbca9ALRkJtz2Z4zLjkYXy1Y1n&pp=sAgC), use the PRDs in this repo, and build everything with Claude Code. You have everything you need right here.