Small updates to AGENTS.md to reflect recent changes

Author: leroyvn

AGENTS.md

@@ -1,14 +1,20 @@
-# CLAUDE.md
+# AGENTS.md
 
-This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+This file provides guidance to coding agents when working with code in this repository.
+
+To use it with Claude Code, simply run the following command in a terminal:
+
+```bash
+echo "@AGENTS.md" > CLAUDE.md
+```
 
 ## Development Commands
 
 This project uses uv as the package manager and taskipy for task automation.
 
 **Setup:**
 ```bash
-uv sync --locked --all-extras --dev
+uv sync --dev --all-extras
 ```
 
 **Testing:**
@@ -47,20 +53,27 @@ xarray-validate is a lightweight validation library for xarray DataArrays and Da
 - `SchemaError`: Custom exception for validation failures
 
 **Validation Components (`components.py`):**
-- `DTypeSchema`: Validates NumPy data types
+- `DTypeSchema`: Validates NumPy data types (supports dtype generics and multiple matches)
 - `DimsSchema`: Validates dimensions (with ordered/unordered support via `ordered` parameter)
 - `ShapeSchema`: Validates array shapes (supports wildcards with None)
 - `NameSchema`: Validates names
 - `ChunksSchema`: Validates dask chunks (bool or dict specification)
 - `ArrayTypeSchema`: Validates underlying array types
-- `AttrSchema`/`AttrsSchema`: Validates attributes
+- `AttrSchema`/`AttrsSchema`: Validates attributes (supports pattern matching for keys and values)
+- `UnitsSchema`: Validates physical units using Pint (requires pint optional dependency)
 
 **Pattern Matching Utilities (`_match.py`):**
 - `_is_regex_pattern()`: Checks if a key is a regex pattern (enclosed in curly braces)
 - `_is_glob_pattern()`: Checks if a key is a glob pattern (contains * or ?)
 - `_is_pattern_key()`: Checks if a key is any kind of pattern
 - `_pattern_to_regex()`: Converts glob or regex patterns to compiled regex objects
-- Used for pattern-based matching in coordinate and data variable keys
+- Used for pattern-based matching in coordinate keys, data variable keys, and attribute keys/values
+
+**Unit Validation Utilities (`units.py`):**
+- `set_registry()`: Set the global Pint unit registry
+- `get_registry()`: Get the current Pint unit registry
+- `parse()`: Parse unit strings with error handling
+- Provides utilities for validating physical units in attributes
 
 **High-Level Schemas:**
 - `DataArraySchema` (`dataarray.py`): Combines all validation components for xarray.DataArray objects
@@ -84,18 +97,25 @@ xarray-validate is a lightweight validation library for xarray DataArrays and Da
 ## Testing Structure
 
 Tests are organized by component:
-- `tests/test_components.py`: Tests for validation components
+- `tests/test_components.py`: Tests for validation components (including unit validation)
 - `tests/test_dataarray.py`: Tests for DataArray schema validation
 - `tests/test_dataset.py`: Tests for Dataset schema validation
-- `tests/conftest.py`: Shared test fixtures
+- `tests/test_lazy_validation.py`: Tests for lazy validation mode
+- `tests/test_match.py`: Tests for pattern matching utilities
+- `tests/test_yaml_examples.py`: Tests for YAML schema loading
+- `conftest.py`: Shared test fixtures (located in project root)
 
-The project targets >90% test coverage and uses pytest with coverage reporting.
+The project uses pytest with xdoctest for docstring testing and coverage reporting.
 
 ## Dependencies and Compatibility
 
 **Core dependencies:** attrs, numpy, xarray
-**Optional dependencies:** dask (for chunk validation), ruamel-yaml (for YAML support)
-**Python support:** 3.8 through 3.13
+**Optional dependencies:**
+- dask (for chunk validation)
+- ruamel-yaml (for YAML support)
+- pint (for unit validation)
+
+**Python support:** 3.8 through 3.14
 **Build system:** hatchling with uv as package manager
 
 ## Code Style
@@ -109,11 +129,14 @@ The project targets >90% test coverage and uses pytest with coverage reporting.
 
 **Module Structure:**
 - `base.py`: Base classes and core abstractions
-- `components.py`: Individual validation components
-- `_match.py`: Pattern matching utilities (shared by dataarray.py and dataset.py)
+- `components.py`: Individual validation components (including UnitsSchema)
+- `_match.py`: Pattern matching utilities (used by dataarray.py, dataset.py, and components.py)
+- `units.py`: Unit validation utilities using Pint
 - `dataarray.py`: DataArray and Coordinates schemas
 - `dataset.py`: Dataset schema
 - `types.py`: Type definitions
+- `testing.py`: Testing utilities
+- `converters.py`: Type converters for attrs
 
 **Key Principles:**
 - Shared utilities are factored into dedicated modules (e.g., `_match.py`)

docs/index.rst

@@ -77,7 +77,7 @@ Development installation:
 
 .. code:: shell
 
-    uv sync --all-groups --all-extras
+    uv sync --dev --all-extras
 
 .. toctree::
     :maxdepth: 2