Examples¶

Start with examples that produce and read real data locally. The examples are intentionally small, but they exercise Artigraph's central model: artifacts, producers, fingerprints, snapshots, and storage metadata.

Snapshot-aware Pandas + Parquet¶

The flagship local example builds partitioned transaction data with Pandas, writes Parquet files, and uses Artigraph snapshot metadata to read a coherent version back.

Run the example

Spend¶

The smaller spend example uses JSON files and Python values. It is useful when you want the fewest moving parts:

uv run python docs/examples/spend/demo.py

Or build and run it in Docker:

docker build --file docs/examples/spend/Dockerfile --tag artigraph/example-spend .
docker run --rm artigraph/example-spend

Ecosystem interop¶

Artigraph-managed files are ordinary files. The difference is that Artigraph also records enough metadata for tools outside Artigraph to know which files belong to a snapshot.

Read about DuckDB interop