Quickstart

Get started with Portolan in minutes. Browse existing catalogs, then publish your own data.

// STEP 1

Browse a catalog

Portolan catalogs are live on the web. Open the Portolan Browser to explore published datasets — view metadata, preview maps, and inspect data tables. Everything you see is served directly from cloud-native files on object storage, with zero servers.

Query with DuckDB

Any GeoParquet file in a Portolan catalog can be queried directly with DuckDB — no download needed. Just point at the URL.

duckdb

$ duckdb

SELECT * FROM read_parquet(

'https://my-catalog.s3.amazonaws.com/parcels.parquet'

) WHERE area_sqm > 1000 LIMIT 10;

// STEP 2

Publish your data

Two paths to get your data into a Portolan catalog.

With the CLI

Install the portolan-cli, then init, add, convert, and push.

install · zsh

$ uv tool install portolan-cli

my-data · zsh

# Initialize a new catalog

$ cd my-data && portolan init

# Add your data files

$ portolan add .

✓ found 12 files (parcels.shp, ortho.tif, roads.shp ...)

# Convert to cloud-native formats

$ portolan check --fix

→ parcels.shp → GeoParquet

→ ortho.tif → COG

→ roads.shp → GeoParquet

✓ STAC catalog generated

# Push to cloud storage

$ portolan push s3://my-bucket/catalog

✓ synced 12 assets to s3://my-bucket/catalog

▸ https://my-bucket.s3.amazonaws.com/catalog.json

done · 0:32 elapsed

With Claude

If you have Claude Code, install the Portolan skill and point Claude at your data. It handles format conversion, metadata generation, and publishing.

install skill · zsh

# In Claude Code, install the Portolan skill

$ claude install portolan-sdi/portolan-cli

claude code

$ claude

> /portolan-cli publish the shapefiles in ./gov-data