lancecli — inspect and explore Lance datasets from the terminal

[itzhakstern]

Hi everyone,

I built lancecli, an open-source command-line tool for inspecting Lance datasets without writing a Python script each time.

It lets you quickly check dataset metadata and schemas, preview and filter rows, fetch specific row positions, and export data as CSV or JSONL.

Quick start

pip install lancecli

# Check dataset metadata
lancecli stat /path/to/dataset.lance

# Preview a few rows
lancecli show /path/to/dataset.lance -n 5

# Fetch specific row positions
lancecli take /path/to/dataset.lance --indices 0,10,100

For wide datasets, you can select just the columns you need and apply a filter:

lancecli show /path/to/dataset.lance \
  -c id,name \
  --filter "id > 100" \
  -n 10

Reads go through pylance: column projection and filters are passed to the Lance scanner, and fetching specific row positions uses Lance’s native take. The tool supports local datasets and S3/S3-compatible storage.

The repository includes a small example dataset and worked command examples.

GitHub: https://github.com/itzhakstern/lancecli
PyPI: https://pypi.org/project/lancecli/

I’d love to hear from others working with Lance: which dataset inspection tasks would you find most useful in a CLI? If you try it on your own datasets, feedback and bug reports are welcome.

[jja725]

we have lance-tools which serve the same purpose, maybe it's better to contribute to that

[dentiny]

FYI I sometimes use DuckDB lance extension for CLI-based query.