How kuljotsb/copilotfullstackapp works
Project Type and Purpose
• This is a full‑stack web application (“Custom Copilot Full Stack App”) that demonstrates a proof‑of‑concept of the Copilot Stack.
• Frontend is React/Vite; backend is a .NET 8 Web API augmented by the Semantic Kernel SDK; it orchestrates calls to Azure OpenAI, Azure AI Search (RAG), and various plugins.
• An Azure Function listens to a Storage Queue to persist chat history into Cosmos DB.
File Structure
Top‑level directories:
• frontend/frontend – React app (components, CSS, index.html, vite.config.js)
• backend – .NET C# Web API (Program.cs, controllers, Dtos, plugin code)
– backend/AISearch – scripts & JSONs to provision AI Search (data ingestion, index, skillset)
– backend/pluginFunctions & backend/plugins – native and prompt‑template plugins loaded by Semantic Kernel
• azureFunctions/StorageQueueFunction – Azure Function (queue trigger) that writes to Cosmos DB
• infrastructure_setup – PowerShell script (deploy.ps1) to provision Azure infra (storage, function, search, cosmos, OpenAI)
• solution file copilot_stack_app.sln
Extract Additional Insights from README
• Architecture uses: React front‑end, .NET Web API, Semantic Kernel as orchestration layer, Azure OpenAI for LLM, Azure Storage (blobs for AI Search docs, queues for chat events), Azure AI Search, Cosmos DB for persistence and audit, Bing Searcher Plugin, Microsoft Graph Plugin.
• Provides step‑by‑step local run instructions, infra deployment, AI Search provisioning, Function run locally.