Chunked N-D arrays with compression and cloud storage. NumPy-style indexing. Backends: local, S3, GCS, ZIP, memory. Dask/Xarray integration for parallel and labeled computation. For lineage use lamindb; for labeled arrays use xarray.
SkillScore95Maintained
Automatic
◆Data Engineer pack
Last verified 8/23/2026
What it changes
Passed 20 of 21 test cases with the skill, and 17 of 21 without it.
Without skill82%
With skill95%
Checks it passed that the model alone missed
Defines time dimension chunk size as 1.
Specifies setting shuffle='shuffle' on the codec.
Recommends GzipCodec for maximum compression ratio.
3 fixed18 passed either way1 still failing0 broke
Pass-rate lift
+14 pp
Tokens
+325%
Agent turns
0%
Response time
-24%
Model gemini-3.6-flashMeasured 8/23/2026v121 of 22 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 2 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 2
Prompt
We are configuring BloscCodec in Python Zarr for an array storing 32-bit floating-point values to improve storage compression. Many adjacent values are similar. Which setting on BloscCodec reorders byte representations prior to compression to increase compression ratio?
What was checked, without and with the skill
Checked
Without skill
With skill
Specifies setting shuffle='shuffle' on the codec.
Missed
Passed
case-04 · gemini-3.6-flash · 8/23/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md531 linessingle file
Zarr Python — Chunked N-D Arrays
Overview
Zarr is a Python library for storing large N-dimensional arrays with chunking, compression, and parallel I/O. It provides NumPy-compatible indexing with pluggable storage backends (local, cloud, in-memory), making it the standard format for cloud-native scientific data pipelines.
When to Use
Storing arrays too large for memory with chunked access (out-of-core computing)
Cloud-native data workflows with S3 or GCS storage backends
Parallel read/write with Dask for large-scale computation
Hierarchical data organization (groups of named arrays with metadata)
Converting between formats (HDF5 → Zarr, NetCDF → Zarr)
Appending time-series data incrementally without rewriting
For labeled, coordinate-aware arrays (time, lat, lon), use xarray with Zarr backend instead
For data management, lineage, and ontology validation, use lamindb (which uses Zarr as a storage format)
Prerequisites
pip install zarr
# Cloud storage support
pip install s3fs # Amazon S3
pip install gcsfs # Google Cloud Storage
Requires Python 3.11+.
Quick Start
import zarr
import numpy as np
# Create a chunked, compressed 2D array
z = zarr.create_array(
store="data/my_array.zarr",
shape=(10000, 10000),
chunks=(1000, 1000),
dtype="f4"