How nishilfaldu/easy-rag works
Project type and purpose
• This is a full-stack web application built on Next.js (React) that lets users create and interact with chatbots.
• It uses Convex as a serverless backend (data modeling, authentication, functions), integrates with S3 for file storage, and external embedding/AI services for document ingestion.
• It also packages a distributable “chatbot-sdk” for embedding the chat widget in other sites.
File structure
• Top-level directories:
• app/ – Next.js “app router” pages, layouts, server- and client-components, split into marketing and main routes.
• src/components, src/app/_components, src/components/ui – reusable React UI.
• src/actions & src/lib – client-side wrappers for calling Convex, S3, and utility functions.
• convex/ – Convex serverless backend:
– schema.ts (data model), auth.config.ts (security), functions.ts, http.ts (API entry points), bots.ts, messages.ts, users.ts, documents.ts
– ingest/ – ingestion pipeline (db.ts, embed.ts, load.ts) calling external AI embedding services.
– _generated/ – Convex client & server code.
• public/ & dist/ – static assets and built chatbot-sdk for CDN inclusion.
• config files – next.config.mjs, tailwind.config.ts, webpack.config.js, tsconfig.json.
README for insights
• Standard Next.js boilerplate; nothing on architecture beyond noting Next.js.
• Runtime: “npm run dev” on localhost:3000.
Elements
• Blue for UI layer
• Green for serverless/backend layer
• Orange for external services/storage
• Annotations for the ingestion pipeline as a sub-component under Convex.