### Setup Pixi Environment

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Clone the repository and set up the development environment using pixi. This includes installing dependencies and activating the development shell.

```bash
# Clone and setup environment
git clone <repository>
cd mcp/
pixi install
```

--------------------------------

### Install Dependencies

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/development.md

Install project dependencies using pixi. Ensure pixi is installed first. Then, install pre-commit hooks.

```shell
pixi install -a
pixi run pre-commit-install
```

--------------------------------

### Install and Run ngff-zarr-mcp with uvx

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Install the uvx package and then run the ngff-zarr-mcp server directly from PyPI. This is the recommended quick installation method.

```bash
pip install uvx

uvx ngff-zarr-mcp
```

--------------------------------

### Handle Installation Permission Issues

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Address permission issues during installation by using user installation with pip or creating and activating a virtual environment. This ensures that packages can be installed without system-level conflicts.

```bash
# Use user installation
pip install --user uvx

# Or create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install ngff-zarr-mcp
```

--------------------------------

### Development Installation with pixi

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Commands for installing pixi, cloning the repository, setting up the development environment, and running development servers or tests.

```bash
# Install pixi if not already installed
curl -fsSL https://pixi.sh/install.sh | bash

# Clone and setup environment
git clone <repository>
cd mcp/
pixi install

# Development environment (includes all dev tools)
pixi shell -e dev

# Run development server
pixi run dev-server

# Run tests and checks
pixi run test
pixi run lint
pixi run typecheck
```

--------------------------------

### Development Installation with pip

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Commands for cloning the repository and installing the package in development mode using pip, followed by instructions to run the server.

```bash
# Clone and install in development mode
git clone <repository>
cd mcp/
pip install -e ".[all]"

# Run the server
ngff-zarr-mcp
```

--------------------------------

### Install Pre-commit Hooks (Bash)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/CONTRIBUTING.md

Install pre-commit hooks for the Python package. Ensure you are in the 'py' directory.

```bash
cd py && pixi run pre-commit-install
```

--------------------------------

### Install ngff-zarr with 'all' Optional Dependencies

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/installation.md

An alternative method to install ngff-zarr with all optional dependencies using the shorthand '[all]'.

```shell
pip install "ngff-zarr[all]"
```

--------------------------------

### Install ngff-zarr TypeScript Package

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/README.md

Installation instructions for the TypeScript package in Deno, Node.js/npm, and browser environments.

```bash
# Deno
import * as ngffZarr from "jsr:@fideus-labs/ngff-zarr";

# Node.js / npm

```

```bash
npm install @fideus-labs/ngff-zarr

# Browser (CDN)

```

```bash
import * as ngffZarr from "https://esm.sh/@fideus-labs/ngff-zarr";

```

--------------------------------

### Install ngff-zarr with CLI

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/quick_start.md

Install the ngff-zarr library with command-line interface support. This is required for using the `ngff-zarr` command.

```shell
pip install "ngff-zarr[cli]"
```

--------------------------------

### Install ngff-zarr-mcp in Zed (uvx)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Add this configuration to your Zed `settings.json` to use uvx for installing ngff-zarr-mcp.

```json
{
  "context_servers": {
    "ngff-zarr": {
      "command": "uvx",
      "args": ["ngff-zarr-mcp"]
    }
  }
}
```

--------------------------------

### Install ngff-zarr-mcp in Claude Desktop (direct)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Add this configuration to your Claude Desktop `claude_desktop_config.json` file for direct installation of ngff-zarr-mcp.

```json
{
  "mcpServers": {
    "ngff-zarr": {
      "command": "ngff-zarr-mcp"
    }
  }
}
```

--------------------------------

### Install ngff-zarr for Node.js/npm

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Install the ngff-zarr package using npm for Node.js environments.

```bash
npm install @fideus-labs/ngff-zarr
```

```typescript
import * as ngffZarr from "@fideus-labs/ngff-zarr";
```

--------------------------------

### Conventional Commit Examples (Bash)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/CONTRIBUTING.md

Examples of commit messages following the Conventional Commits format for different types and scopes.

```bash
feat(py): add support for zarr v3 sharding
fix(mcp): handle missing metadata gracefully
docs: update installation instructions
chore(ts): update dependencies
```

--------------------------------

### Install NGFF-Zarr with JSR (Deno/Node.js)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Import the library using JSR for Deno or modern Node.js environments.

```typescript
import * as ngffZarr from "jsr:@fideus-labs/ngff-zarr";
```

```bash
npx jsr add @fideus-labs/ngff-zarr
```

```typescript
import * as ngffZarr from "@fideus-labs/ngff-zarr";
```

--------------------------------

### Install vkfft for GPU-accelerated ITK

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/methods.md

Install the itk-vkfft package to enable GPU-accelerated, FFT-based filtering for ITK-based downsampling methods. This is applied by default after installation.

```sh
pip install itk-vkfft
```

--------------------------------

### Install ngff-zarr with Tensorstore Support

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/python.md

Install the 'ngff-zarr' package with the 'tensorstore' extra to enable writing OME-Zarr stores using Tensorstore for potentially better performance.

```shell
pip install "ngff-zarr[tensorstore]"
```

--------------------------------

### Conventional Commit Examples

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Examples of valid commit messages following the conventional commits specification for different scopes and types.

```bash
feat(py): add support for RFC-9 OME-Zarr format
fix(ts): resolve memory leak in multiscale generation
docs: update installation instructions
chore(mcp): update dependencies
```

--------------------------------

### Install TIFF Support

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/tiff.md

Install ngff-zarr with CLI optional dependencies to enable TIFF support. This includes the 'tifffile' library.

```shell
pip install 'ngff-zarr[cli]'
```

--------------------------------

### Read OME-Zarr Basic Example

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Perform a basic read operation on an OME-Zarr file using the fromNgffZarr function.

```typescript
// Basic reading
const ms = await fromNgffZarr("data.ome.zarr");

// With validation
const validatedMs = await fromNgffZarr("data.ome.zarr", {
  validate: true,
  version: "0.5",
});

// From URL
const remoteMs = await fromNgffZarr("https://example.com/data.ome.zarr");

```

--------------------------------

### Install ngff-zarr with All Optional Dependencies

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/installation.md

Install ngff-zarr along with all available optional dependencies for comprehensive functionality, including CLI tools, dask-image, ITK, TensorStore, and validation.

```shell
pip install "ngff-zarr[cli,dask-image,itk,tensorstore,validate]"
```

--------------------------------

### Install ngff-zarr via pip

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/installation.md

Use this command to install the ngff-zarr library for general use on your system.

```shell
pip install ngff-zarr
```

--------------------------------

### Install ngff-zarr with Validation Support

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/python.md

Install the 'ngff-zarr' package with the 'validate' extra to enable OME-Zarr metadata validation.

```shell
pip install "ngff-zarr[validate]"
```

--------------------------------

### Create and Write OME-Zarr Files

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Construct OME-Zarr data from scratch, including image data, axes, scales, and metadata, then write it to a Zarr store. This example creates a simple grayscale image.

```typescript
import {
  createAxis,
  createDataset,
  createMetadata,
  createMultiscales,
  createNgffImage,
  toNgffZarr,
} from "@fideus-labs/ngff-zarr";

// Create image data (256x256 grayscale)
const data = new Uint8Array(256 * 256);
for (let i = 0; i < data.length; i++) {
  data[i] = Math.floor(Math.random() * 256);
}

// Create an NgffImage with metadata
const image = createNgffImage(
  data.buffer,
  [256, 256],           // shape
  "uint8",              // dtype
  ["y", "x"],          // dimension names
  { y: 1.0, x: 1.0 },  // scale (pixel spacing)
  { y: 0.0, x: 0.0 }   // translation (origin)
);

// Create OME-Zarr metadata
const axes = [
  createAxis("y", "space", "micrometer"),
  createAxis("x", "space", "micrometer"),
];

const datasets = [
  createDataset(
    "0",                           // path
    [1.0, 1.0],                   // scale
    [0.0, 0.0]                    // translation
  ),
];

const metadata = createMetadata(axes, datasets, "my_image");

// Create multiscales container
const multiscales = createMultiscales([image], metadata);

// Write to OME-Zarr
await toNgffZarr("output.ome.zarr", multiscales);
```

--------------------------------

### Configure ngff-zarr MCP Server in VS Code (pip install)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Configuration for ngff-zarr MCP server in VS Code using pip install. This method installs the package and then runs the server directly.

```json
"mcp": {
  "servers": {
    "ngff-zarr": {
      "command": "python",
      "args": ["-c", "import subprocess; subprocess.run(['pip', 'install', 'ngff-zarr-mcp']); import ngff_zarr_mcp.server; ngff_zarr_mcp.server.main()"]
    }
  }
}
```

--------------------------------

### Install NGFF-Zarr for Browser (CDN)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Include the library directly in the browser using a CDN.

```html
<script type="module">
  import * as ngffZarr from "https://esm.sh/@fideus-labs/ngff-zarr";

  // Your code here
</script>
```

--------------------------------

### Convert Image to OME-Zarr with Metadata via MCP Client

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Example demonstrating how to convert a NIfTI-GZ image to OME-Zarr, including specifying dimensions, scale, units, and name.

```python
result = await convert_images_to_ome_zarr(
    input_paths=["image.nii.gz"],
    output_path="brain.ome.zarr",
    dims=["z", "y", "x"],
    scale={"z": 2.0, "y": 0.5, "x": 0.5},
    units={"z": "micrometer", "y": "micrometer", "x": "micrometer"},
    name="Brain MRI",
    scale_factors=[2, 4]
)
```

--------------------------------

### OpenCode MCP Server Configuration

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/mcp.md

Example JSON configuration for integrating the ngff-zarr-mcp server with OpenCode. Adjust 'command' and 'environment' for other agents.

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "ngff-zarr": {
      "type": "local",
      "command": ["uvx", "ngff-zarr-mcp"],
      "enabled": true
    }
  }
}
```

--------------------------------

### Install ngff-zarr for Deno

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Import the ngff-zarr library for use in Deno projects.

```typescript
import * as ngffZarr from "@fideus-labs/ngff-zarr";
```

--------------------------------

### Install fsspec and s3fs

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/faq.md

Required packages for reading from S3, GCS, and other remote storage using fsspec.

```bash
pip install fsspec s3fs
```

--------------------------------

### Changelog Filtering Example (Markdown)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Example of a changelog entry with a clickable GitHub commit link. This format is generated by the custom Commitizen plugin.

```markdown
- **py**: add feature ([abc1234](https://github.com/fideus-labs/ngff-zarr/commit/abc1234...))
```

--------------------------------

### Debug Browser Tests with Deno

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Use this command to start a debugging session for browser tests. Ensure you are in the project root.

```bash
deno task test:browser:debug
```

--------------------------------

### Convert Image to OME-Zarr with itk-wasm

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

This example demonstrates loading an image using itk-wasm, converting it to NgffImage format, generating a multiscale pyramid, and writing it to an OME-Zarr store.

```typescript
import {
  itkImageToNgffImage,
  toMultiscales,
  toNgffZarr,
} from "@fideus-labs/ngff-zarr";

// Load image using itk-wasm
const itkImage = await loadImageFromFile("input.png");

// Convert to NgffImage
const ngffImage = await itkImageToNgffImage(itkImage);

// Generate pyramid
const multiscales = await toMultiscales(ngffImage);

// Write to OME-Zarr
await toNgffZarr("output.ome.zarr", multiscales);
```

--------------------------------

### Install ngff-zarr-mcp in Claude Code (uvx)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Run this command in Claude Code to add ngff-zarr using uvx.

```sh
claude mcp add ngff-zarr -- uvx ngff-zarr-mcp
```

--------------------------------

### Install ngff-zarr with Traditional Tools

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Install ngff-zarr in development mode using pip and traditional tools. This method is an alternative to pixi for managing dependencies and running development tasks.

```bash
# Clone and install in development mode
git clone <repository>
cd mcp/
pip install -e ".[all]"

# Run tests
pytest

# Lint code
black .
ruff check .
```

--------------------------------

### Python Import Pattern for Core Functions (Python)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Example of importing core functions from the ngff_zarr library, emphasizing function-based imports over classes.

```python
from ngff_zarr import from_ngff_zarr, to_ngff_zarr, to_multiscales
```

--------------------------------

### Access metadata

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/lif.md

This example shows how to access and print metadata from a LIF image after converting it to an NgffImage object.

```APIDOC
## Access metadata

This example shows how to access and print metadata from a LIF image after converting it to an NgffImage object.

```python
from liffile import LifFile
from ngff_zarr import lif_to_ngff_image

with LifFile("microscopy_data.lif") as lif:
    for i, lif_image in enumerate(lif.images):
        ngff_image = lif_to_ngff_image(lif_image)

        print(f"\nSeries {i}: {lif_image.path}")
        print(f"  Dimensions: {ngff_image.dims}")
        print(f"  Shape: {ngff_image.data.shape}")
        print(f"  Scale (pixel size): {ngff_image.scale}")
        print(f"  Translation (origin): {ngff_image.translation}")

        # Access timestamps if available
        if ngff_image.attrs and "timestamps" in ngff_image.attrs:
            print(f"  Timestamps: {ngff_image.attrs['timestamps']}")
```
```

--------------------------------

### Install ngff-zarr in Browser with Pyodide

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/installation.md

Install the ngff-zarr library within a browser environment using Pyodide, suitable for web-based applications or interactive consoles like the Pyodide REPL or JupyterLite.

```python
import micropip
await micropip.install('ngff-zarr')
```

--------------------------------

### Python Development Commands (Bash)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/CONTRIBUTING.md

Commands for developing the Python package, including installation, testing, linting, and building documentation.

```bash
cd py
pixi install                    # Install dependencies
pixi run test                   # Run test suite
pixi run lint                   # Run linting
pixi run build-docs             # Build documentation
```

--------------------------------

### MCP Server Development Commands (Bash)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/CONTRIBUTING.md

Commands for developing the MCP server package, including installation, testing, type checking, formatting, and linting.

```bash
cd mcp
pixi install                    # Install dependencies
pixi run test                   # Run tests
pixi run typecheck              # Type checking (mypy)
pixi run format                 # Format code
pixi run lint                   # Run linting
```

--------------------------------

### Complete Drug Screening Workflow Setup (v0.4)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/hcs.md

Imports necessary modules for a drug screening workflow using ngff-zarr, including HCSPlate and to_hcs_zarr, and specific metadata components for version 0.4.

```python
import ngff_zarr as nz
import numpy as np
from ngff_zarr import NgffImage, to_multiscales
from ngff_zarr.hcs import HCSPlate, to_hcs_zarr
from ngff_zarr.v04.zarr_metadata import *

```

--------------------------------

### TypeScript Import Pattern for Core Functions (TypeScript)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Example of importing core functions in TypeScript, using function-based exports and JSR imports.

```typescript
import { fromNgffZarr, toNgffZarr } from "./io/from_ngff_zarr.ts";
```

--------------------------------

### Check ngff-zarr-mcp Help

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Verify that the ngff-zarr-mcp server starts correctly by checking its help message. This is part of troubleshooting general MCP client errors.

```bash
uvx ngff-zarr-mcp --help
```

--------------------------------

### Working with Different Store Types

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Examples of using ngff-zarr with various storage backends, including in-memory maps, HTTP/HTTPS fetch stores, and file system stores. Note that file system stores are only available in Node.js or Deno environments.

```typescript
import * as zarr from "zarrita";
import { fromNgffZarr } from "@fideus-labs/ngff-zarr";

// Memory store
const memoryStore = new Map<string, Uint8Array>();
const ms1 = await fromNgffZarr(memoryStore);

// Fetch store (HTTP/HTTPS)
const fetchStore = new zarr.FetchStore("https://example.com/data.ome.zarr");
const ms2 = await fromNgffZarr(fetchStore);

// File system store (Node.js/Deno only)
import { FileSystemStore } from "@zarrita/storage";
const fsStore = new FileSystemStore("/path/to/data.ome.zarr");
const ms3 = await fromNgffZarr(fsStore);
```

--------------------------------

### Convert Single Image to OME-Zarr via MCP Client

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Example of using the MCP client to convert a single TIFF image to OME-Zarr format with specified scale factors and compression.

```python
# Through MCP client, the agent can:
result = await convert_images_to_ome_zarr(
    input_paths=["image.tif"],
    output_path="output.ome.zarr",
    ome_zarr_version="0.4",
    scale_factors=[2, 4, 8],
    method="itkwasm_gaussian",
    compression_codec="zstd"
)
```

--------------------------------

### Configure ngff-zarr MCP Server in Cursor (Direct Python)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Configuration for ngff-zarr MCP server in Cursor using direct Python execution. This method involves installing the package and then running the server.

```json
{
  "mcpServers": {
    "ngff-zarr": {
      "command": "python",
      "args": ["-m", "pip", "install", "ngff-zarr-mcp", "&&", "ngff-zarr-mcp"]
    }
  }
}
```

--------------------------------

### Create and Write HCS Plate (v0.4)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/hcs.md

A complete example for creating an HCS plate structure and writing images to its wells using v0.4 metadata. Ensure all operations use the same version.

```python
import ngff_zarr as nz
import numpy as np
from ngff_zarr.hcs import HCSPlate, to_hcs_zarr
from ngff_zarr.v04.zarr_metadata import *

# IMPORTANT: Define version once - all plate and well operations must use the same version
ome_zarr_version = "0.4"

# Create plate layout
columns = [PlateColumn(name="1"), PlateColumn(name="2")]
rows = [PlateRow(name="A"), PlateRow(name="B")]
wells = [
    PlateWell(path="A/1", rowIndex=0, columnIndex=0),
    PlateWell(path="A/2", rowIndex=0, columnIndex=1),
    PlateWell(path="B/1", rowIndex=1, columnIndex=0),
    PlateWell(path="B/2", rowIndex=1, columnIndex=1),
]

plate_metadata = Plate(
    columns=columns, rows=rows, wells=wells,
    name="Example Plate v0.4", field_count=1,
    version=ome_zarr_version
)

# Create synthetic image data
data = np.random.randint(0, 255, size=(1, 1, 10, 256, 256), dtype=np.uint8)
ngff_image = nz.NgffImage(
    data=data, dims=["t", "c", "z", "y", "x"],
    scale={"t": 1.0, "c": 1.0, "z": 0.5, "y": 0.325, "x": 0.325},
    translation={"t": 0.0, "c": 0.0, "z": 0.0, "y": 0.0, "x": 0.0}
)
multiscales = nz.to_multiscales(ngff_image)

# First create the plate structure
hcs_plate = HCSPlate(store="my_plate_v04.ome.zarr", plate_metadata=plate_metadata)
to_hcs_zarr(hcs_plate, "my_plate_v04.ome.zarr")

# Write images to each well (all must use same version as plate)
for well in wells:
    row_name = rows[well.rowIndex].name
    col_name = columns[well.columnIndex].name

    nz.write_hcs_well_image(
        store="my_plate_v04.ome.zarr",
        multiscales=multiscales,
        plate_metadata=plate_metadata,
        row_name=row_name,
        column_name=col_name,
        field_index=0,
        version=ome_zarr_version,  # Must match plate version
    )

# Verify the result
plate = nz.from_hcs_zarr("my_plate_v04.ome.zarr")
print(f"Created v0.4 plate with {len(plate.wells)} wells")
```

--------------------------------

### Build Documentation

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/development.md

Build and update the project documentation if necessary.

```shell
pixi run dev-docs
```

--------------------------------

### Install cuCIM for GPU-accelerated ITKWASM

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/methods.md

Install the cuCIM package to enable NVIDIA CUDA GPU acceleration for ITKWASM downsampling methods. GPU-accelerated filtering is applied by default after installation.

```sh
pip install itkwasm-downsample-cucim
```

--------------------------------

### Build npm package and run browser tests

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

A convenience command to first build the npm package and then run the browser tests.

```bash
deno task test:browser:npm
```

--------------------------------

### Set Up Plate Structure with to_hcs_zarr

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/hcs.md

Create the HCS plate structure on disk using `to_hcs_zarr`. Ensure the `HCSPlate` object is initialized with the store path and plate metadata.

```python
import ngff_zarr as nz
from ngff_zarr.hcs import HCSPlate, to_hcs_zarr

# Create the HCS plate structure
hcs_plate = HCSPlate(store="my_screen.ome.zarr", plate_metadata=plate_metadata)
to_hcs_zarr(hcs_plate, "my_screen.ome.zarr")
```

--------------------------------

### Zarr Store Handling (Python)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Python code demonstrating how to open Zarr stores, automatically detecting v2 or v3 and using the appropriate store type.

```python
from zarr.storage import DirectoryStore, LocalStore  # v2 vs v3
store = zarr.open.v2() if zarr_v2 else zarr.open()
```

--------------------------------

### Initialize Results Table

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/test/browser-npm/omero-benchmark.html

Populates the results table with rows for each benchmark configuration, showing pending status and initial memory estimates.

```javascript
function initResultsTable() {
  resultsBody.innerHTML = "";
  BENCHMARK_CONFIGS.forEach((config) => {
    const elements = config.shape.reduce((a, b) => a * b, 1);
    const memoryMB = ((elements * 4) / (1024 * 1024)).toFixed(1);
    const row = document.createElement("tr");
    row.id = `row-${config.name}`;
    row.innerHTML = `
      <td>${config.name}</td>
      <td class="number">${elements.toLocaleString()}</td>
      <td>${memoryMB} MB</td>
      <td><span class="status-badge status-pending">Pending</span></td>
      <td class="number">-</td>
      <td class="number">-</td>
      <td class="number">-</td>
      <td class="number">-</td>
    `;
    resultsBody.appendChild(row);
  });
}
```

--------------------------------

### Run tests with Deno

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Execute the project's test suite using Deno. Ensure all tests pass to verify code integrity.

```bash
deno test --allow-all
```

--------------------------------

### Run browser tests with UI

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Launch browser tests with an interactive UI for easier debugging and observation.

```bash
deno task test:browser:ui
```

--------------------------------

### Run Interactive Commit Helper (Bash)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/CONTRIBUTING.md

Use the interactive CLI to help create compliant commit messages. Navigate to the respective package directory first.

```bash
cd py && pixi run commit
# or from other directories:
cd ts && pixi run commit
cd mcp && pixi run commit
```

--------------------------------

### Install ITK dependencies for ngff-zarr

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/methods.md

Install the necessary dependencies for using ITK-based downsampling methods with ngff-zarr. This includes native binary builds.

```sh
pip install "ngff-zarr[itk]"
```

--------------------------------

### Run All Pre-commit Hooks (Bash)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Command to run all pre-commit hooks, including Ruff formatting and checking, from the 'py/' directory. Essential for AI agents to ensure code compliance.

```bash
pixi run --as-is lint      # from py/ directory — runs all pre-commit hooks
```

--------------------------------

### Install ngff-zarr CLI

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/cli.md

Install the command line interface for ngff-zarr using pip. This command includes the necessary dependencies for CLI usage.

```shell
pip install 'ngff-zarr[cli]'
```

--------------------------------

### Run Test Suite

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/development.md

Execute the project's test suite using the pixi run command.

```shell
pixi run test
```

--------------------------------

### Install DASK_IMAGE dependencies for ngff-zarr

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/methods.md

Install the required dependencies for using DASK_IMAGE-based downsampling methods with ngff-zarr. This enables implementations based on dask-image and scipy.

```sh
pip install "ngff-zarr[dask-image]"
```

--------------------------------

### Run all tests including browser tests

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Execute the complete test suite, encompassing both standard and browser-based tests.

```bash
deno task test:all
```

--------------------------------

### Run TypeScript Tests and Build Commands

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Commands for running Deno test suite, linting, formatting, building, browser compatibility tests, and type checking for the TypeScript package.

```bash
cd ts && pixi run --as-is test          # Deno test suite
cd ts && pixi run --as-is lint          # Deno lint
cd ts && pixi run --as-is fmt           # Deno format
cd ts && pixi run --as-is build         # Full build pipeline
cd ts && pixi run --as-is test:browser  # Browser compatibility tests
cd ts && pixi run --as-is check         # Type checking
```

--------------------------------

### Write OME-Zarr Basic Example

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Write an NgffMultiscales object to an OME-Zarr file using the toNgffZarr function.

```typescript
// Basic writing
await toNgffZarr("output.ome.zarr", multiscales);

// With version specification
await toNgffZarr("output.ome.zarr", multiscales, { version: "0.5" });

// With sharding (v0.5)
await toNgffZarr("output.ome.zarr", multiscales, {
  version: "0.5",
  chunksPerShard: { z: 2, y: 2, x: 2 },
});

```

--------------------------------

### Install ngff-zarr-mcp in Claude Code (pip)

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Run this command in Claude Code to add ngff-zarr using pip.

```sh
claude mcp add ngff-zarr -- python -m pip install ngff-zarr-mcp && ngff-zarr-mcp
```

--------------------------------

### Install ngff-zarr-mcp in OpenCode

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Add this configuration to your OpenCode settings to enable ngff-zarr tools. Restart OpenCode after adding the configuration.

```json
{
  "mcp": {
    "ngff-zarr": {
      "type": "local",
      "command": [
        "python",
        "-c",
        "import subprocess; subprocess.run(['pip', 'install', 'ngff-zarr-mcp']); import ngff_zarr_mcp.server; ngff_zarr_mcp.server.main()"
      ],
      "enabled": true
    }
  }
}
```

--------------------------------

### Convert OME-Zarr Directory to .ozx File

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/python.md

Read an existing OME-Zarr directory store and write it as a portable .ozx file. This creates a single-file archive for easy distribution.

```python
>>> # Read from directory store
>>> multiscales = nz.from_ngff_zarr('cthead1.ome.zarr')
>>>
>>> # Write as .ozx file
>>> nz.to_ngff_zarr('cthead1.ozx', multiscales)
```

--------------------------------

### Build npm package with Deno

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/ts/README.md

Create an npm-compatible package from the TypeScript code using Deno's build task.

```bash
deno task build:npm
```

--------------------------------

### Run MCP Server Tests and Development Commands

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/AGENTS.md

Commands for testing, type checking, formatting, and running the MCP server in development mode for the MCP server package.

```bash
cd mcp && pixi run --as-is test         # Run MCP tests
cd mcp && pixi run --as-is typecheck    # mypy type checking
cd mcp && pixi run --as-is format       # Format code
cd mcp && pixi run --as-is dev          # Run MCP server in dev mode
```

--------------------------------

### RFC-4 Metadata Format Example

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/rfc4.md

Shows the JSON structure for OME-NGFF metadata when RFC-4 is enabled, including the 'orientation' field for spatial axes.

```json
{
  "axes": [
    {
      "name": "z",
      "type": "space",
      "unit": "micrometer",
      "orientation": {
        "type": "anatomical",
        "value": "inferior-to-superior"
      }
    },
    {
      "name": "y",
      "type": "space",
      "unit": "micrometer",
      "orientation": {
        "type": "anatomical",
        "value": "anterior-to-posterior"
      }
    },
    {
      "name": "x",
      "type": "space",
      "unit": "micrometer",
      "orientation": {
        "type": "anatomical",
        "value": "right-to-left"
      }
    }
  ]
}
```

--------------------------------

### Generate Multiscale Pyramids

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/typescript.md

Create multiscale image pyramids with specified downsampling factors and methods. This example uses Gaussian downsampling and specifies chunking.

```typescript
import {
  createNgffImage,
  toMultiscales,
  toNgffZarr,
  Methods,
} from "@fideus-labs/ngff-zarr";

// Create base image
const data = new Uint8Array(512 * 512);
const image = createNgffImage(
  data.buffer,
  [512, 512],
  "uint8",
  ["y", "x"],
  { y: 1.0, x: 1.0 },
  { y: 0.0, x: 0.0 }
);

// Generate multiscale pyramid with 2x and 4x downsampling
const multiscales = await toMultiscales(image, {
  scaleFactors: [2, 4],
  method: Methods.ITKWASM_GAUSSIAN,
  chunks: 128,
});

// Write the pyramid
await toNgffZarr("pyramid.ome.zarr", multiscales);
```

--------------------------------

### Create NgffImage with LPS orientation

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/docs/rfc4.md

Demonstrates how to create an NgffImage using the LPS convenience constant for axes orientations. This image is then converted to multiscales and written to a zarr store with RFC-4 enabled.

```python
import ngff_zarr as nz
import dask.array as da

# Create image data
data = da.zeros((100, 100, 100))

# Using the LPS convenience constant
ngff_image = nz.NgffImage(
    data=data,
    dims=("z", "y", "x"),
    scale={"x": 1.0, "y": 1.0, "z": 1.0},
    translation={"x": 0.0, "y": 0.0, "z": 0.0},
    axes_orientations=nz.LPS
)

# Convert to multiscales and write with RFC-4 enabled
multiscales = nz.to_multiscales(ngff_image)
nz.to_ngff_zarr(
    store="output.ome.zarr",
    multiscales=multiscales,
    enabled_rfcs=[4]
)
```

--------------------------------

### Install ngff-zarr-mcp in Augment Code UI

Source: https://github.com/fideus-labs/ngff-zarr/blob/main/mcp/README.md

Use the Augment Code UI to add ngff-zarr-mcp by entering the command `uvx ngff-zarr-mcp` and naming the MCP 'ngff-zarr'.

```text
uvx ngff-zarr-mcp
```