Kineviz (formerly GraphXR) is Kineviz's graph visualization and analytics platform. Some product surfaces — the Google Cloud Marketplace listing, the
graphxr.kineviz.comportal, and screenshots in this repo — still show the former name.
Explore Spanner property graphs visually in Kineviz. Two ways in — pick the one that matches why you're here.
→ connect/ — point Kineviz at an existing property graph. About ten
minutes, no demo data.
→ demos/ — a worked example that builds a graph in your own instance,
verifies it, and hands you questions to ask.
| Demo | What it shows | Level | Time | Cost |
|---|---|---|---|---|
fraud-rings |
Build a property graph of synthetic P2P payments in Spanner and find the accounts that share a device and move money between themselves. | beginner | 15 min | free |
Generated from each demo's demo.yaml — edit that, not this table.
git clone https://github.com/Kineviz/spanner-kineviz-examples
cd spanner-kineviz-examples
./gxr list
./gxr up fraud-rings- A Kineviz account — sign up. Kineviz Desktop is free for individual use, forever; the app requires sign-in, so do this first.
- Kineviz Desktop v0.17.1+ — Windows, macOS, or Linux. ~600 MB installed, 16 GB RAM.
- An existing Spanner instance, plus
gcloudand Python 3.9+.
Spanner Graph is GA — no preview terms, and no edition or reservation requirement for GQL. If you can read the database, you can query the graph.
Spanner's model is the opposite of BigQuery's, and it is the thing that catches people out.
| BigQuery | per byte scanned. Idle costs nothing |
| Spanner | provisioned capacity, billed continuously whether you query or not |
So the expensive object is the instance, not the data. Every demo here creates a database inside an instance you already run — storage only, a few MB — and teardown drops only that database, never the instance, which may hold other people's work.
No instance? A free trial instance is 90 days at no charge and covers everything in this repo.
./gxr down <demo> # drops the demo's database; leaves your instance alonePoint Claude Code, Codex, or Cursor at this repo — AGENTS.md tells it how to
stand a demo up. Three things it won't do, by design: create your Kineviz account, install
Desktop, or sign in for you.
| Path | |
|---|---|
connect/ |
Connect Kineviz to your own graph — standalone |
demos/ |
Worked examples |
AGENTS.md |
Agent entry point |
gxr |
list · preflight · up · verify · down · doctor |
tools/ |
check_contract.py, selftest.sh, preview.sh, gen_index.py |
bigquery-kineviz-examples— BigQuery Graphalloydb-kineviz-examples— AlloyDB and PostgreSQLkineviz-desktop— the appgraphxr-database-proxy— zero-trust middleware with a Spanner driver, if a credential must not sit on a laptop
See CONTRIBUTING.md. Demo suggestions welcome — the dataset must be public or synthetic.