Mini-Agent: Log File Debugger (Next.js + FastAPI + Postgres)

Upload a log file → backend flags lines and returns a concise summary with suggested fixes.

Stack

• Frontend: Next.js (deployed on Vercel)
• Backend: FastAPI (separate service)
• DB: Postgres (stores sessions/messages; optional for log mode)
• LLM: OpenAI (server-side only)

Run locally (no Docker)

1. Frontend install

npm install

2. Backend deps

pip3 install --break-system-packages -r backend/requirements.txt

3. Environment

cp .env.example .env
# set OPENAI_API_KEY in .env (optional but recommended)

4. Start backend (SQLite by default)

bash backend/run.sh
# backend: http://localhost:8000

5. Start frontend

export NEXT_PUBLIC_API_URL=http://localhost:8000
npm run dev
# app: http://localhost:3000

Run with Docker Compose

OPENAI_API_KEY=sk-... docker compose up --build
# backend: http://localhost:8000  |  postgres: localhost:5432 (app/app)

Apply SQL (optional; app auto-creates SQLite tables, Postgres uses SQL):

psql postgresql://app:app@localhost:5432/app -f db/migrations/0001_init.sql

Deploy

• Frontend (Vercel):

  • Framework Preset: Next.js
  • Root Directory: ./
  • Build Command: npm run build
  • Env: NEXT_PUBLIC_API_URL=https://YOUR-BACKEND

• Backend (Railway/Render/Fly):

  • Entrypoint: bash backend/run.sh
  • Env: DATABASE_URL, OPENAI_API_KEY, CORS_ORIGINS=https://YOUR-VERCEL

Endpoints

• POST /chat/log-analyze (multipart file: file) → { flagged[], analysis }
• POST /chat/sessions → create session
• GET /chat/sessions/{id} → session with messages
• POST /chat/sessions/{id}/message → add message and get assistant reply