NOVA

Agentic AI learning and academic success platform for students, teachers, and admins.

What it does

NOVA centers on grounded student-facing AI workflows that read real app data, propose structured outputs, and require approval before mutating state.

Current student agent surfaces:

• Academic Success Agent at /student/agent
• Career Roadmap at /student/career
• Research Assistant at /student/research
• Project Mentor at /student/project-mentor

Core agent flows

Academic Success Agent

Reads timetable, attendance, and assignments. Proposes study plans. Student approves or rejects. Approved plans can be completed or skipped. Skipped sessions can trigger reschedule proposals.

Career Roadmap

Reads student profile, academic history, and prior roadmap memory. Proposes milestone-based career roadmap. Student approves or rejects pending roadmap.

Research Assistant

Finds research sources for topic, summarizes response in chat, and lets student save session into ResearchNote records.

Project Mentor

Reads existing projects, proposes milestone plans for new project work, and supports milestone updates. New project plans now enter pending approval flow before becoming active.

Tech stack

• Next.js 15 App Router
• React 19
• TypeScript
• Tailwind CSS
• Radix UI + shadcn/ui
• MongoDB + Mongoose
• NextAuth.js v4
• OpenRouter / OpenAI-compatible tool calling
• Flask ML service in apps/ml-api

Repo layout

apps/
  web/
    app/
      api/
        student/
          agent/
          career-agent/
          research-agent/
          project-mentor/
          orchestrator/
      student/
        agent/
        career/
        research/
        project-mentor/
    lib/
      academic-agent.ts
      career-agent.ts
      research-agent.ts
      project-agent.ts
      agent-tools.ts
      models.ts
  ml-api/
docs/

Setup

Prerequisites

• Node.js >= 18
• pnpm
• running MongoDB instance
• OpenRouter-compatible API key for agent flows

Environment

Copy root example file into web app env file:

cp .env.example apps/web/.env.local

Minimum required values:

• MONGODB_URI
• MONGODB_DB_NAME
• NEXTAUTH_SECRET
• NEXTAUTH_URL
• OPENROUTER_API_KEY

Common optional values:

• OPENROUTER_API_BASE
• OPENROUTER_MODEL
• PERFORMANCE_PREDICTOR_API_URL

Run locally

Install dependencies:

pnpm -C apps/web install

Seed demo data:

pnpm -C apps/web seed:demo

Start web app:

pnpm -C apps/web dev

Start ML API separately if needed:

cd apps/ml-api
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python app.py

Important student API routes

• POST /api/student/agent
• POST /api/student/agent/action
• POST /api/student/agent/session
• POST /api/student/career-agent
• POST /api/student/career-agent/action
• POST /api/student/research-agent
• POST /api/student/research-agent/save
• POST /api/student/project-mentor
• POST /api/student/project-mentor/action

All student mutation routes derive acting student from getServerSession. Client does not supply student id.

Data models used by agent flows

• StudyPlan
• AgentAction
• CareerRoadmap
• ResearchNote
• Project

Notes

• Research Assistant save path reuses existing persistence logic through saveResearchNote.
• Career Roadmap and Project Mentor now both have dedicated approval routes instead of relying only on orchestrator path.
• Sidebar groups student AI entries under single AI Agents section.

Verification

Latest local verification after agent UI/API updates:

• pnpm -C apps/web exec tsc --noEmit passes
• pnpm -C apps/web lint does not pass repo-wide because of pre-existing unrelated lint errors, including apps/web/app/page.tsx

Docs

• Product/design direction: docs/Design.md
• project rules: docs/Rules.md
• phase notes: docs/Phases.md