CognitionFlow

Not another knowledge base. A local-first cognition system that ingests information, understands it, and grows with you.

CognitionFlow is a local-first project for people who feel overwhelmed by fast-moving technical information and fragmented personal notes.

It combines:

• multi-source ingestion: RSS, arXiv, GitHub Trending, and more
• AI understanding: summarization, tagging, clustering, and future graph extraction
• local knowledge: connect with an Obsidian vault or plain Markdown notes
• continuous discovery: recommend what is new and why it matters to your existing knowledge

Architecture

┌─────────────────────────────────────────────────────────────┐
│                        Frontend (Web)                      │
│             React + Vite + TypeScript dashboard            │
└─────────────────────────────────────────────────────────────┘
                              │ HTTP
┌─────────────────────────────────────────────────────────────┐
│                     Backend API (FastAPI)                  │
│  feeds • items • summaries • recommendations • settings    │
└─────────────────────────────────────────────────────────────┘
         │                        │                       │
         │                        │                       │
   Ingestion services        Knowledge sync         Scheduler/jobs
 RSS / arXiv / GitHub        Obsidian / Markdown    Daily refresh
         │                        │                       │
         └─────────────── SQLite / local storage ───────────┘

Monorepo layout

CognitionFlow/
├── backend/
│   ├── app/
│   │   ├── api/
│   │   ├── core/
│   │   ├── models/
│   │   └── services/
│   ├── tests/
│   ├── pyproject.toml
│   └── .env.example
├── frontend/
│   ├── src/
│   ├── package.json
│   └── vite.config.ts
├── docs/
│   └── roadmap.md
├── scripts/
│   └── bootstrap.sh
├── docker-compose.yml
└── README.md

Quick start

1) Backend

cd backend
python -m venv .venv
source .venv/bin/activate
pip install -e .
cp .env.example .env
uvicorn app.main:app --reload --port 8000

2) Frontend

cd frontend
npm install
npm run dev

Frontend will run on http://localhost:5173 and talk to the backend at http://localhost:8000.

First milestone

• Monorepo scaffold
• FastAPI health/feed endpoints
• React dashboard skeleton
• pluggable ingestion service interface
• RSS ingestion MVP
• arXiv ingestion
• GitHub Trending ingestion
• Obsidian sync
• daily brief generation
• recommendation ranking

Product principle

CognitionFlow is not trying to become another passive note app.

It should help answer:

1. What changed today that matters to me?
2. How does it connect to what I already know?
3. What should I read next?

License

MIT