Invana

Invana is an open-source graph intelligence platform that turns structured knowledge graphs into interactive decision simulation environments.

Development setup

Prerequisites

• Python 3.14+
• Node.js 22+
• uv (Python package manager)
• pnpm (Node package manager)
• Docker & Docker Compose

Quick start

# 1. Clone and install dependencies + pre-commit hooks
git clone https://github.com/invana/invana.git
cd invana
make setup

# 2. Start postgres, neo4j, the engine, and studio (hot-reload dev containers)
docker compose up -d

# 3. Run migrations + bootstrap the default root superuser
#    (admin / hi@invana.local / change_me_please — change the password after first login)
make engine-init

Studio runs at http://localhost:8300, the engine at http://localhost:8200. Log in with hi@invana.local / change_me_please. Run make help to see all available commands.

docker compose up -d starts only the core stack — postgres, neo4j, engine, studio. Neo4j is the only graph database that starts by default; the extra ones (Memgraph, JanusGraph, ArcadeDB) and the observability stack (HyperDX) are opt-in via Docker Compose profiles. Prefer running the engine and studio directly on your host? Use make dev instead of the engine/studio containers (stop those two, or don't start them, first — same infra either way). See CONTRIBUTING.md for the full guide — infra profiles, environment variables, optional dependency extras, testing, changesets, commit conventions, and the pull-request process.