Welcome to the ultimate repository of modular, production-grade AI agent architectures. This cookbook contains practical, standalone recipes demonstrating advanced cognitive design patterns using LangChain and LangGraph.

Rather than simple, toy examples, each sub-repository houses a self-contained, real-world agent pipeline with structured state handling, resilient fallbacks, and multi-model integrations.

This ecosystem is split into three highly specialized agent blueprints. You can navigate into any sub-directory to find dedicated setup manuals and visual workflow graphs:

langchain-cookbook/
├── full_agent/          # Hierarchical Deep Agent with Task Tracking
├── RAG-agent/           # Dual-Approach RAG: Tool-driven vs Prompt Middleware
├── SQL-agent/           # Autonomous Text-to-SQL with Self-Correction
├── LICENSE              # Repository open-source licensing terms
└── README.md            # Global workspace registry (This file)

• Core Stack: LangChain, LangGraph
• Design Pattern: Supervisor-Worker Hierarchy
• Key Feature: Solves token bloat and context pollution by utilizing an autonomous supervisor agent that dynamically coordinates tasks, tracks progress with a live digital TODO checklist, and spins up isolated worker frames for background research.

• Core Stack: ChromaDB, PyPDF
• Design Pattern: Tool Ingestion vs Prompt Middleware Injection
• Key Feature: Directly compares two distinct RAG architectures. One allows the LLM to choose when to deploy a search tool, while the other intercepts user input via custom @dynamic_prompt middleware to seamlessly inject PDF vector context before inference.

• Core Stack: LangGraph, PostgreSQL
• Design Pattern: Self-Correction & Fuzzy Fallback Loops
• Key Feature: A robust Text-to-SQL sales assistant. Features an automated syntax-checking engine that fixes raw queries before execution, grades retrieved results, and leverages regex fuzzy fallbacks (LIKE) to ensure the end application interface never gets blank data screens.

Every recipe in this cookbook is completely modular and self-contained. To run any agent locally:

1. Clone this repository to your machine.
2. Open your terminal and change into the specific recipe directory:

   cd SQL-agent # Or full_agent / RAG-agent