Mini Research Team

Mini Research Team is a small full-stack project that demonstrates an AI-powered research workflow. It includes a Next.js client UI (client/) and a NestJS backend (server/) which runs research "workflows" (summarization, splitting, ranking, cross-checking) and stores documents in MongoDB.

Tech stack

• Frontend: Next.js 15 + React 19, Tailwind CSS (client)
• Backend: NestJS 11, Mongoose (server)
• AI: Google Generative AI client packages present in server dependencies
• State & data fetching: Redux Toolkit + RTK Query (client)

Key features

• Ask a research question from the UI and run a multi-step workflow that produces a final answer and a trace of intermediate steps.
• Upload documents for the knowledge base (documents stored via the docs module).
• Trace viewer: inspect JSON trace of workflow steps.

Repository layout

• client/ — Next.js app (UI, upload page, question input, results)
• server/ — NestJS API (workflow orchestration, docs, upload, trace endpoints)
• server/src/workflow/ — services implementing the research workflow (splitter, summarizer, ranker, crosschecker, etc.)

Getting started Prerequisites

• Node.js (v18+ recommended)
• npm or yarn
• MongoDB (local or hosted) — used by the server
• Google Cloud credentials / API key for the Generative AI client (if you plan to run workflows that call Google APIs)

Local development

1. Start the backend

   cd server
   npm install
   # set environment variables (see below), then:
   npm run start:dev

2. Start the frontend

   cd client
   npm install
   npm run dev

The frontend runs on the Next.js dev server (see client/package.json scripts). The backend default port is 4000 (see server/src/main.ts).

Environment

• MONGODB_URI — MongoDB connection string (defaults to mongodb://localhost:27017/mini_research in AppModule).
• PORT — backend listen port (defaults to 4000).
• Google AI credentials — configure according to the Google Generative AI client you use (environment variable or service account JSON). The server currently includes @google/genai / @google/generative-ai in dependencies.

Useful scripts

• Frontend (client/package.json)

  • dev — Next.js dev server
  • build — build production assets
  • start — start built Next app

• Backend (server/package.json)

  • start:dev — NestJS dev server with watch
  • start — start (compiled) server
  • build — compile TypeScript to dist
  • seed — run ts-node src/seed/seed.ts to seed sample data
  • test — run Jest tests

Important files

• Server bootstrap: server/src/main.ts
• Server module & DB config: server/src/app.module.ts
• Client main UI: client/app/page.tsx
• Workflow services: server/src/workflow/ (splitter, summarizer, ranker, crosschecker)

Notes & next steps

• The repository is scaffolded with a clear separation of concerns: UI (Next.js) and API (NestJS). To enable actual calls to Google Generative AI, make sure your Google credentials are configured in the environment before starting the server.
• Consider adding a top-level dev script or a docker-compose to run both services together.

Contributing

• Open an issue or PR when you want to add features or fix bugs. Follow existing code style and tests.

License

• See individual packages. The server package.json is marked UNLICENSED by default; add a license file if you plan to open-source this project.