llms.txt

# Context Engineering for Multi-Agent Systems

> In 21st century AI, LLMs are the agents, and the MAS is the environment they operate in.

This repository provides a production-ready, glass-box framework for building domain-agnostic Multi-Agent Systems (MAS). It moves beyond prompting into Context Engineering, utilizing the Model Context Protocol (MCP) to orchestrate specialized agents within a verifiable "Context Engine" architecture.

## Core Metadata
- **Author:** Denis Rothman
- **Key Concepts:** Context Engineering, Multi-Agent Systems (MAS), Semantic Blueprints, High-Fidelity RAG, Model Context Protocol (MCP)
- **Primary Model Support:** Optimized for GPT-5.1 (adaptive reasoning) and backward compatible with GPT-4 series
- **Status:** Updated February 2026

## Key Features
- **Universal Context Engine:** A domain-agnostic core that runs cross-domain use cases (e.g., Legal and Marketing) without changing code.
- **High-Fidelity RAG:** Research agent with automated input sanitization and source-verifiable citations.
- **Token Analytics:** Integrated tracking of Input/Output tokens and cost-efficiency dashboards.
- **Sovereign Architecture:** A "Glass Box" approach providing full transparency into agent reasoning traces.

## Technical Stack
- **Languages:** Python 3.10+
- **Protocols:** Model Context Protocol (MCP)
- **Vector DB:** Pinecone
- **LLM APIs:** OpenAI (latest standards)
- **Dependencies:** Tenacity (>=9.0.0), Tiktoken, FastAPI, OpenAI

## Project Structure
- `/Chapter10/`: Universal Context Engine (Batch & UI versions)
- `/commons/engine/`: Core logic including `engine.py`, `agents.py`, and `registry.py`
- `/Chapter08-09/`: Specialized implementations for Legal and Strategic Marketing domains
- `CHANGELOG.md`: Detailed history of GPT-5.1 upgrades and framework evolution

## Getting Started
1. Obtain OpenAI and Pinecone API keys.
2. Use provided Google Colab or Kaggle badges in the README for instant deployment.
3. Reference `engine.py` for the main execution logic and `agents.py` for specialist agent definitions.

---
*For a detailed narrative of the software evolution from 1980s Latent Semantic Structures to the 2026 Agentic Era, see the media/index.html timeline.*