diff --git a/pyproject.toml b/pyproject.toml
index 0f1a8c3..df2cbbc 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -161,9 +161,9 @@ select = [
     "YTT", # flake8-2020
 ]
 ignore = [
-    "C408", # Unnecessary {obj_type} call (rewrite as a literal)
-    "B905", # zip() without an explicit strict= parameter
-    "C901", # {name} is too complex ({complexity} > {max_complexity})
+    "C408",   # Unnecessary {obj_type} call (rewrite as a literal)
+    "B905",   # zip() without an explicit strict= parameter
+    "C901",   # {name} is too complex ({complexity} > {max_complexity})
     "COM812",
     # "CPY001", # Missing copyright notice at top of file
     "D100",    # Missing docstring in public module
@@ -211,6 +211,7 @@ combine-as-imports = true
 
 [tool.ruff.lint.per-file-ignores]
 "__init__.py" = ["E402", "I001"]
+"tests/test_decorators.py" = ["PLR2004"]
 
 [tool.pyright]
 reportUnusedCallResult = false
diff --git a/src/jinjarope/decorators.py b/src/jinjarope/decorators.py
new file mode 100644
index 0000000..7850e75
--- /dev/null
+++ b/src/jinjarope/decorators.py
@@ -0,0 +1,127 @@
+from __future__ import annotations
+
+from functools import wraps
+from typing import TYPE_CHECKING, Any, ParamSpec, TypeVar
+
+import upath
+
+
+if TYPE_CHECKING:
+    from collections.abc import Callable
+
+
+P = ParamSpec("P")
+R = TypeVar("R")
+
+
+def cache_with_transforms(
+    *,
+    arg_transformers: dict[int, Callable[[Any], Any]] | None = None,
+    kwarg_transformers: dict[str, Callable[[Any], Any]] | None = None,
+) -> Callable[[Callable[P, R]], Callable[P, R]]:
+    """A caching decorator with transformation functions for args and kwargs.
+
+    Can be used to make specific args / kwargs hashable.
+    Also adds cache and cache_info objects to the decorated function.
+
+    Args:
+        arg_transformers: Dict mapping positional args indices to transformer functions
+        kwarg_transformers: Dict mapping kwargs names to transformer functions
+
+    Returns:
+        A decorator function that caches results based on transformed arguments
+    """
+    arg_transformers = arg_transformers or {}
+    kwarg_transformers = kwarg_transformers or {}
+
+    def decorator(func: Callable[P, R]) -> Callable[P, R]:
+        cache: dict[tuple[Any, ...], R] = {}
+
+        @wraps(func)
+        def wrapper(*args: P.args, **kwargs: P.kwargs) -> R:
+            # Transform positional arguments
+            transformed_args = tuple(
+                arg_transformers.get(i, lambda x: x)(arg) for i, arg in enumerate(args)
+            )
+
+            # Transform keyword arguments
+            transformed_kwargs = {
+                key: kwarg_transformers.get(key, lambda x: x)(value)
+                for key, value in sorted(kwargs.items())
+            }
+
+            # Create cache key from transformed arguments
+            cache_key = (transformed_args, tuple(transformed_kwargs.items()))
+
+            if cache_key not in cache:
+                cache[cache_key] = func(*args, **kwargs)
+            return cache[cache_key]
+
+        def cache_info() -> dict[str, int]:
+            """Return information about cache hits and size."""
+            return {"cache_size": len(cache)}
+
+        wrapper.cache_info = cache_info  # type: ignore
+        wrapper.cache = cache  # type: ignore
+
+        return wrapper
+
+    return decorator
+
+
+# def path_cache(
+#     maxsize: int | None = None,
+#     typed: bool = False,
+#     args_index: int = 0,
+# ) -> Callable[[Callable[P, R]], Callable[P, R]]:
+#     """Decorator that caches function results and normalizes PathLike arguments.
+
+#     Caches the first argument by default. Can be changed via args_index argument.
+#     Also adds cache_info and cache_clear methods to the decorated function.
+
+#     Args:
+#         maxsize: Maximum size of the cache. If None, cache is unbounded.
+#         typed: If True, arguments of different types are cached separately.
+#         args_index: Index of the argument to cache. Default is 0.
+
+#     Returns:
+#         Decorated function with caching capability.
+#     """
+
+#     def decorator(func: Callable[P, R]) -> Callable[P, R]:
+#         # Create cached version of the original function
+#         cached_func = lru_cache(maxsize=maxsize, typed=typed)(func)
+
+#         @wraps(func)
+#         def wrapper(*args: P.args, **kwargs: P.kwargs) -> R:
+#             # Convert first argument if it's PathLike
+#             if args and isinstance(args[args_index], str | PathLike):
+#                 normalized_args = (upath.UPath(args[0]).resolve(),) + args[1:]
+#                 return cached_func(*normalized_args, **kwargs)  # type: ignore[arg-type]
+#             return cached_func(*args, **kwargs)  # type: ignore[arg-type]
+
+#         # Add cache_info and cache_clear methods to the wrapper
+#         wrapper.cache_info = cached_func.cache_info  # type: ignore
+#         wrapper.cache_clear = cached_func.cache_clear  # type: ignore
+
+#         return wrapper
+
+#     return decorator
+
+
+if __name__ == "__main__":
+
+    @cache_with_transforms(arg_transformers={0: lambda p: upath.UPath(p).resolve()})
+    def read_file_content(filepath: str | upath.UPath) -> str:
+        """Read and return the content of a file."""
+        with upath.UPath(filepath).open() as f:
+            return f.read()
+
+    # These calls will use the same cache entry
+    content1 = read_file_content("pyproject.toml")
+    content1 = read_file_content("mkdocs.yml")
+    content2 = read_file_content(upath.UPath("pyproject.toml"))
+    content3 = read_file_content(upath.UPath("./pyproject.toml").absolute())
+
+    # Check cache statistics
+    print(read_file_content.cache_info())  # type: ignore
diff --git a/tests/test_decorators.py b/tests/test_decorators.py
new file mode 100644
index 0000000..5b7d6c6
--- /dev/null
+++ b/tests/test_decorators.py
@@ -0,0 +1,176 @@
+# mypy: disable-error-code="attr-defined"
+from typing import Any
+
+import pytest
+
+from jinjarope.decorators import cache_with_transforms
+
+
+def test_basic_caching() -> None:
+    """Test basic function caching without transformers."""
+    call_count = 0
+
+    @cache_with_transforms()
+    def add(x: int, y: int) -> int:
+        nonlocal call_count
+        call_count += 1
+        return x + y
+
+    assert add(1, 2) == 3
+    assert add(1, 2) == 3
+    assert call_count == 1
+    assert add.cache_info()["cache_size"] == 1
+
+
+def test_arg_transformer() -> None:
+    """Test caching with argument transformers."""
+
+    @cache_with_transforms(arg_transformers={0: lambda x: x.lower()})
+    def greet(name: str) -> str:
+        return f"Hello, {name}!"
+
+    assert greet("John") == "Hello, John!"
+    assert greet("JOHN") == "Hello, John!"  # Should hit cache, previous word cached
+    assert greet.cache_info()["cache_size"] == 1
+
+
+def test_kwarg_transformer() -> None:
+    """Test caching with keyword argument transformers."""
+
+    @cache_with_transforms(kwarg_transformers={"items": tuple})
+    def process_list(*, items: list[int]) -> int:
+        return sum(items)
+
+    assert process_list(items=[1, 2, 3]) == 6
+    assert process_list(items=[1, 2, 3]) == 6  # Should hit cache
+    assert process_list.cache_info()["cache_size"] == 1
+
+
+def test_unhashable_args() -> None:
+    """Test caching with unhashable arguments."""
+
+    @cache_with_transforms(arg_transformers={0: tuple})
+    def process_list(lst: list[int]) -> int:
+        return sum(lst)
+
+    assert process_list([1, 2, 3]) == 6
+    assert process_list([1, 2, 3]) == 6  # Should hit cache
+    assert process_list.cache_info()["cache_size"] == 1
+
+
+def test_mixed_args_kwargs() -> None:
+    """Test caching with both positional and keyword arguments."""
+
+    @cache_with_transforms(
+        arg_transformers={0: str.lower}, kwarg_transformers={"items": tuple}
+    )
+    def process_data(prefix: str, *, items: list[int]) -> str:
+        return f"{prefix}: {sum(items)}"
+
+    assert process_data("Sum", items=[1, 2, 3]) == "Sum: 6"
+    assert process_data("SUM", items=[1, 2, 3]) == "Sum: 6"  # Should hit cache
+    assert process_data.cache_info()["cache_size"] == 1
+
+
+def test_multiple_calls_different_args() -> None:
+    """Test caching behavior with different arguments."""
+    call_count = 0
+
+    @cache_with_transforms()
+    def add(x: int, y: int) -> int:
+        nonlocal call_count
+        call_count += 1
+        return x + y
+
+    assert add(1, 2) == 3
+    assert add(2, 3) == 5
+    assert add(1, 2) == 3  # Should hit cache
+    assert call_count == 2
+    assert add.cache_info()["cache_size"] == 2
+
+
+def test_none_values() -> None:
+    """Test caching behavior with None values."""
+
+    @cache_with_transforms()
+    def process_optional(x: Any) -> str:
+        return str(x)
+
+    assert process_optional(None) == "None"
+    assert process_optional(None) == "None"  # Should hit cache
+    assert process_optional.cache_info()["cache_size"] == 1
+
+
+def test_empty_transformers() -> None:
+    """Test that decorator works with empty transformers."""
+
+    @cache_with_transforms(arg_transformers={}, kwarg_transformers={})
+    def identity(x: int) -> int:
+        return x
+
+    assert identity(1) == 1
+    assert identity.cache_info()["cache_size"] == 1
+
+
+def test_cache_persistence() -> None:
+    """Test that cache persists between calls."""
+
+    @cache_with_transforms()
+    def expensive_operation(x: int) -> int:
+        return x**2
+
+    assert expensive_operation(2) == 4
+    initial_cache = expensive_operation.cache
+    assert expensive_operation(2) == 4
+    assert expensive_operation.cache is initial_cache
+
+
+def test_different_kwarg_orders() -> None:
+    """Test that different keyword argument orders produce same cache key."""
+
+    @cache_with_transforms()
+    def process_kwargs(*, a: int, b: int) -> int:
+        return a + b
+
+    assert process_kwargs(a=1, b=2) == 3
+    assert process_kwargs(b=2, a=1) == 3  # Should hit cache
+    assert process_kwargs.cache_info()["cache_size"] == 1
+
+
+def test_complex_transformers() -> None:
+    """Test with more complex transformer functions."""
+
+    def complex_transform(x: list[Any]) -> tuple[Any, ...]:
+        return tuple(sorted(x))
+
+    @cache_with_transforms(
+        arg_transformers={0: complex_transform},
+        kwarg_transformers={"data": complex_transform},
+    )
+    def process_lists(lst: list[int], *, data: list[int]) -> int:
+        return sum(lst) + sum(data)
+
+    assert process_lists([3, 1, 2], data=[6, 4, 5]) == 21
+    assert process_lists([2, 3, 1], data=[5, 6, 4]) == 21  # Should hit cache
+    assert process_lists.cache_info()["cache_size"] == 1
+
+
+def test_error_handling() -> None:
+    """Test that errors in the original function are not cached."""
+
+    @cache_with_transforms()
+    def failing_function(x: int) -> int:
+        msg = "Error"
+        raise ValueError(msg)
+
+    with pytest.raises(ValueError):  # noqa: PT011
+        failing_function(1)
+
+    with pytest.raises(ValueError):  # noqa: PT011
+        failing_function(1)  # Should not be cached
+
+    assert failing_function.cache_info()["cache_size"] == 0
+
+
+if __name__ == "__main__":
+    pytest.main([__file__])