How utkarshmohan2003/docsqueeze works
Project type and purpose
• DocSqueeze is a full-stack SaaS application for uploading PDF documents and generating AI-powered summaries.
• Built on Next.js (App Router), it uses Prisma ORM + SQLite for persistence, Clerk for auth, UploadThing for file uploads, and Google Gemini Flash 2.0 for summarization.
File structure
• app/
• Pages & layouts (App Router)
• app/page.tsx (public home)
• app/sign-in, app/sign-up (Clerk auth flows)
• app/(logged-in)/upload & app/(logged-in)/your-summaries (protected pages)
• app/api/uploadthing/route.ts (UploadThing serverless route)
• components/
• Reusable UI: common (header, footer), home (hero, cta, pricing), upload form, summary list
• lib/
• Business logic & integrations:
– geminiAi.ts, langchain.ts, openai.ts (AI model wrappers)
– prisma.ts, queries.ts (DB client & queries)
– prompts.ts, utils.ts
• prisma/
• Schema + migrations (SQLite in dev, can swap to Postgres in prod)
• utils/
• uploadthing.ts (client helper for UploadThing)
• Config & infra files: next.config.ts, middleware.ts, tsconfig.json, schema.sql
README for additional insights
• Confirms the tech stack and feature set
• Highlights modular, scalable architecture
• Mentions deployment target: Vercel / Node server