A modular, production-ready AI assistant backend built with Python, LangChain, OpenAI API, and RAG (Retrieval-Augmented Generation) pipeline.

• 🔍 RAG Pipeline: Embed documents and perform similarity search for context retrieval
• 🤖 LangChain Agent: Chain of tools including Calculator and Google Search
• 💬 Conversational Memory: Maintains conversation history using LangChain's ConversationBufferMemory
• 🧾 Document Ingestion: Support for PDFs, TXT, Markdown, and DOCX files
• 🛠️ Embeddings: OpenAI embeddings for document vectorization
• 🗂️ Vector Store: ChromaDB for fast document retrieval
• 🔑 Environment-based Configuration: Secure API key management
• 📄 Swagger Documentation: Auto-generated API docs
• 🚀 FastAPI Backend: High-performance async API

ContextAgent/ ├── app/ │ ├── main.py # FastAPI app entrypoint │ ├── routes/ │ │ ├── chat.py # Chat endpoints │ │ └── ingest.py # Document upload endpoints │ ├── chains/ │ │ ├── qa_chain.py # RAG + LLM chain │ │ └── agent_chain.py # LangChain agent setup │ ├── tools/ │ │ ├── calculator.py # Custom LangChain tools │ │ └── google_search.py # Web search tool │ ├── memory/ │ │ └── session_memory.py # Conversational memory │ ├── ingest/ │ │ ├── embedder.py # Embedding function │ │ └── vector_store.py # ChromaDB setup │ ├── utils/ │ │ ├── config.py # Environment + settings │ │ └── document_loader.py # Document processing │ └── schemas/ │ └── request_model.py # Pydantic schemas ├── .env.example # Environment variables template ├── requirements.txt # Python dependencies └── README.md # This file 

git clone git https://github.com:webcodelabb/ContextAgent.git cd ContextAgent

pip install -r requirements.txt

Copy the example environment file and add your API keys:

cp env.example .env

Edit .env and add your OpenAI API key:

OPENAI_API_KEY=your_openai_api_key_here OPENAI_MODEL=gpt-4

python -m app.main

The server will start at http://localhost:8000

Ask questions to the AI assistant.

Request Body:

{ "question": "What does this PDF say about climate change?", "history": [ {"role": "user", "content": "Summarize the document"}, {"role": "agent", "content": "Sure, here's the summary..."} ], "use_rag": true, "use_agent": false }

Response:

{ "answer": "The PDF discusses the recent changes in global temperatures and the effects of greenhouse gases...", "sources": ["climate_report_2024.pdf"], "reasoning": null, "metadata": { "model": "gpt-4", "session_id": "default", "documents_retrieved": 3 } }

Get conversation history for a session.

Clear conversation memory for a session.

Get information about available tools.

Get statistics about the chat system.

Upload a document for processing.

Supported formats: PDF, TXT, MD, DOCX

Ingest all supported documents from a directory.

Get statistics about ingested documents.

Clear all ingested documents.

Check system health and configuration.

import requests # Chat with RAG response = requests.post("http://localhost:8000/chat/", json={ "question": "What are the main points in the uploaded documents?", "use_rag": True }) print(response.json()["answer"]) # Chat with Agent response = requests.post("http://localhost:8000/chat/", json={ "question": "What's 15 * 23?", "use_agent": True }) print(response.json()["answer"])

# Simple chat curl -X POST "http://localhost:8000/chat/" \ -H "Content-Type: application/json" \ -d '{"question": "Hello, how are you?"}' # Upload document curl -X POST "http://localhost:8000/ingest/upload" \ -F "file=@document.pdf" # Get system stats curl "http://localhost:8000/chat/stats"

1. Start the server: python -m app.main
2. Open Swagger docs: http://localhost:8000/docs
3. Test endpoints through the interactive interface

# Health check curl http://localhost:8000/health # Upload a test document curl -X POST "http://localhost:8000/ingest/upload" \ -F "file=@test_document.pdf" # Ask a question curl -X POST "http://localhost:8000/chat/" \ -H "Content-Type: application/json" \ -d '{"question": "What is this document about?"}'

1. Document Ingestion: Upload PDFs, TXTs, MDs, DOCXs
2. Text Processing: Split documents into chunks
3. Embedding: Convert text to vectors using OpenAI
4. Storage: Store in ChromaDB vector database
5. Retrieval: Find relevant documents for queries
6. Generation: Generate answers using LLM with context

• Calculator Tool: Perform mathematical calculations
• Google Search Tool: Search the web for current information
• Conversational Memory: Maintains chat history
• Multi-step Reasoning: Chain multiple tools together

• PDF: PyPDF2 for text extraction
• TXT: UTF-8 text files
• MD: Markdown files
• DOCX: Microsoft Word documents
• Chunking: Intelligent text splitting with overlap
• Metadata: Preserves source information

FROM python:3.11-slim WORKDIR /app COPY requirements.txt . RUN pip install -r requirements.txt COPY . . EXPOSE 8000 CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]

# Production environment export OPENAI_API_KEY=your_production_key export OPENAI_MODEL=gpt-4 export HOST=0.0.0.0 export PORT=8000

• Set up proper CORS configuration
• Use environment variables for secrets
• Implement authentication if needed
• Monitor API usage and costs
• Set up logging and monitoring

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

This project is licensed under the MIT License.

• LangChain for the amazing LLM framework
• OpenAI for the GPT models
• ChromaDB for vector storage
• FastAPI for the web framework

Built with ❤️ for the AI community Testing coauthor achievement Fixing coauthor format