AI-powered document review app. Upload a PDF, extract content via a vision LLM, review it with grammar/style/consistency agents, and edit the result in a rich text editor.
- Backend — Python, FastAPI, pydantic-ai, fasta2a (A2A protocol)
- Frontend — React, TypeScript, TipTap, Tailwind CSS
- LLMs — LM Studio (vision: gemma-3-12b, text: gpt-oss-20b)
# Backend
cd backend
cp .env.example .env # edit with your LM Studio URLs
poetry install
poetry run uvicorn app.agents.a2a_app:app --port 9000 &
poetry run uvicorn app.main:app --port 8000 &
# Frontend
cd frontend
npm install
npm run devdocker compose up --buildfrontend (React + TipTap) → server (FastAPI :8000) → agents (A2A :9000)
├── grammar
├── style
├── consistency
├── page_extractor (vision)
└── markdown_fixer