Documentation Maintainer

Documentation Maintainer is an AI-powered system for reviewing, updating, and managing technical documentation. It combines a modern web frontend with an intelligent backend to streamline the process of keeping documentation accurate and up-to-date.

---

Overview

• Natural Language Querying: Ask questions or describe changes in plain English to find relevant documentation sections.
• AI Suggestions: Receive suggested edits, additions, or removals powered by large language models (LLMs).
• Review Workflow: Approve, reject, or edit AI-generated suggestions before saving changes.
• Modern UI: Clean, responsive interface for efficient review and collaboration.

---

Architecture

• Frontend: Built with Next.js and React, providing a side-by-side diff view, markdown rendering, and intuitive controls. See Frontend README for details.
• Backend: FastAPI-based service orchestrating a Retrieval-Augmented Generation (RAG) pipeline using LangChain, ChromaDB, and Pydantic. See Backend README for details.
• Data Storage: Uses ChromaDB to store document chunks and embeddings. See Data Storage and Retrieval.

Backend Architecture Diagram

The diagram above illustrates the core components of the backend system:

• FastAPI Application: Serves as the main entry point, handling API requests from the frontend.
• RAG Pipeline: Implements Retrieval-Augmented Generation using LangChain to find and process relevant documentation chunks.
• ChromaDB: Stores document embeddings and metadata, enabling efficient semantic search and retrieval.
• LLM Integration: Connects to large language models (OpenAI or Google) to generate suggestions and edits based on user queries.
• Pydantic Models: Used for data validation and structured responses throughout the API.

This architecture enables scalable, intelligent document management and seamless integration with the modern frontend.

---

Key Features

• Query → Suggest → Review → Save: Enter a query, get relevant docs and AI suggestions, review changes, and update documentation.
• Diff Highlighting: Visualize changes with color-coded badges and markdown support.
• Flexible Embeddings: Supports Google and OpenAI embedding models for document retrieval.
• Evaluation Methods: Includes ground-truth, unit, negative, and human-in-the-loop testing. See Framework Evaluation.

---

Getting Started

1. Install uv (recommended):

   curl -LsSf https://astral.sh/uv/install.sh | sh

2. Install dependencies:

   uv sync

3. Documentation Markdown Files

   • place the markdown files inside: data/documentation/

4. Set up environment variables:

   • Copy .env.example to .env in fastapi_backend/ and fill in required keys.
   • Input DOC_DIR_PATH as data/documentation in .env

5. Run the backend:

   cd fastapi_backend
   uvicorn routes:app --host 0.0.0.0--reload

6. Run the frontend:

   cd nextjs_frontend
   npm install
   npm run dev

See Deployment and Scaling for more.

---

Documentation

• Frontend Guide
• Backend Guide
• Data Storage & Retrieval
• Framework Evaluation
• Deployment & Scaling