Skip to main content

Build Status PyPI

strict-kwargs

Enforce using keyword arguments where possible.

strict-kwargs is a standalone CLI implemented in Rust.

For example, if we have a function which takes two regular arguments, there are three ways to call it. With this tool, only the form where keyword arguments are used is accepted.

"""Showcase errors when calling a function without naming the arguments."""


def add(a: int, b: int) -> int:
    """Add two numbers."""
    return a + b


add(a=1, b=2)  # OK
add(1, 2)  # strict-kwargs reports this; strict-kwargs check --fix can rewrite it
add(1, b=2)  # strict-kwargs reports this; strict-kwargs check --fix can rewrite it

Why?

  • Like a formatter, such as black or ruff format, this lets you stop discussing whether a particular function call should use keyword arguments.
  • Positional arguments can be fine at first. As more are added, calls can become unclear without anyone stopping to refactor them to keyword arguments.
  • Type checkers give better errors when keyword arguments are used. For example, with positional arguments, you may see, Argument 5 to "add" has incompatible type "str"; expected "int". This requires that you count the arguments to see which one is wrong. With named arguments, you get Argument "e" to "add" has incompatible type "str"; expected "int".

Installation

uv tool install strict-kwargs

or:

pip install strict-kwargs

This is tested on Python 3.11+.

Usage

strict-kwargs check .                         # check a directory
strict-kwargs check --output-format json .    # emit check diagnostics as JSON
strict-kwargs check --output-format github .  # emit GitHub Actions annotations
strict-kwargs check --fix .                   # rewrite positional args in place
strict-kwargs check --diff .                  # preview fixes, write nothing
strict-kwargs check --fix --unsafe-fixes .    # include behavior-changing fixes
strict-kwargs check --python .venv .          # point type resolution at an environment
strict-kwargs check --cache-dir .strict-kwargs-cache .  # enable the diagnostic cache

Exit codes are:

  • 0: clean
  • 1: violations found
  • 2: operational error

Output

  • full, the default check output, writes Ruff-style diagnostics and summaries to stdout.
  • json and github write diagnostics to stdout so machine consumers can read them without mixing in operational messages.
  • Warnings and operational errors are always written to stderr.
  • check --diff writes the unified diff to stdout and its summary to stderr.

Fix behavior

check --fix only rewrites calls whose target parameter names are known unambiguously. Ambiguous calls are counted as declined.

By default, check --fix rewrites:

  • single-signature calls
  • overloaded calls, when one precise overload arm can be selected and the rewritten argument types are precise enough

Synthesized constructors are treated as unsafe fixes because generated constructor models can differ from runtime behaviour when class construction is customized.

  • --unsafe-fixes: include dataclass and NamedTuple constructor calls whose signatures were synthesized from fields.

Python environment

Use --python to point third-party resolution at an interpreter, virtual environment, or sys.prefix.

  • Missing paths are errors.
  • A missing --python path is warned about and ignored.

pre-commit

repos:
  - repo: https://github.com/adamtheturtle/strict-kwargs-pre-commit
    rev: 2026.8.24  # pin to a release tag
    hooks:
      - id: strict-kwargs

Configuration

Configuration lives in pyproject.toml:

[tool.strict_kwargs]
required_version = ">=2026.5.19-post.3"
ignore_names = ["main.func", "builtins.str"]
src = ["src"]
namespace_packages = ["src/airflow/providers"]
extend_exclude = ["generated", "vendor"]
force_exclude = true
cache_dir = ".strict-kwargs-cache"
fix_synthesized_constructors = true
error_on_unused_noqa = true
output_format = "full"  # or "json", "github"

Set required_version to make older or incompatible strict-kwargs binaries fail fast when they read this project configuration. Supported specifiers are exact versions, such as 2026.5.19-post.3, and minimum versions, such as >=2026.5.19-post.3. Use the version reported by strict-kwargs --version.

Ignored functions

Use ignore_names for functions that should still allow positional arguments. This is useful especially for builtins which can look strange with keyword arguments.

For example, str(object=1) is not idiomatic.

Suppressing individual findings

Add a Ruff-style # noqa comment to the line a diagnostic is reported on (the first line of the offending call) to suppress it:

func(1, 2, 3)  # noqa: KW001
  • # noqa: KW001 suppresses only KW001. A directive naming other codes (for example # noqa: E501) leaves the call reported.
  • A bare # noqa suppresses every finding on the line, matching Ruff.
  • Suppressed calls are skipped by --fix too, so a # noqa call is never rewritten.

For a call spanning multiple lines, put the comment on the first line, the line the path:line:col output points at:

func(  # noqa: KW001
    1,
    2,
    3,
)

Finding # noqa comments that are no longer needed

Set error_on_unused_noqa = true (or pass --error-on-unused-noqa) to report a KW002 error for every # noqa: KW001 directive that suppressed nothing:

main.py:3:12: KW002 Unused `noqa` directive (unused: `KW001`)

Only directives that name KW001 explicitly are reported. A bare # noqa, or one naming only other tools' codes, is left alone: strict-kwargs sees only its own rule, so it cannot tell whether such a directive is suppressing someone else's finding.

KW002 errors count towards the exit code of check. They belong to checking rather than fixing: --error-on-unused-noqa cannot be combined with --fix or --diff, and those modes neither report nor remove unused directives even when the rule is enabled in pyproject.toml. Enabling the rule makes suppressed calls cost a full check (including the ty fallback), so a run with many # noqa comments is somewhat slower.

Using # noqa alongside Ruff

If you also run Ruff with RUF100 (unused noqa) enabled, prefer the coded form # noqa: KW001: Ruff leaves a directive whose only codes it does not recognise untouched, but it will remove a bare # noqa it considers unused. To keep KW001 from being stripped when it shares a directive with a Ruff code (for example # noqa: E501, KW001), declare it as an external code:

[tool.ruff.lint]
external = ["KW001"]

Source discovery

Set src to source-code directories that should be:

  • searched for first-party imports
  • stripped when deriving module names

Relative paths are resolved against the project root. For example, src = ["src"] maps src/pkg/mod.py to pkg.mod while preserving the repository root as a fallback source root.

Set namespace_packages to directories that should be treated as namespace packages for module resolution even when they have no __init__.py.

Exclusions

Use extend_exclude to skip generated or vendored Python files during directory runs.

  • Patterns use .gitignore-style matching relative to the project root.
  • By default, exclusions apply to directory traversal only.
  • An explicitly passed file, such as strict-kwargs check generated/api.py, is still checked.
  • Set force_exclude = true to apply exclusions to explicitly passed files too. This is useful when pre-commit passes changed files directly.
  • The built-in skips for dot-directories, venv, and __pycache__ remain enabled.

Cache

Set cache_dir to enable the persistent diagnostic cache for strict-kwargs checks. Relative cache_dir values in pyproject.toml are resolved against the project root.

The cache location precedence is:

  1. --cache-dir
  2. [tool.strict_kwargs].cache_dir
  3. STRICT_KWARGS_CACHE_DIR

If none are set, the cache is disabled.

Fix defaults

Set fix_synthesized_constructors = true to make strict-kwargs check --fix include dataclass and NamedTuple constructor rewrites without passing --unsafe-fixes each time.

To find the name of a function to ignore, set the following configuration:

[tool.strict_kwargs]
debug = true

Then run strict-kwargs check and look for the debug output.

Comparison with mypy-strict-kwargs

mypy-strict-kwargs is a mypy plugin that enforces the same rule during type checking.

Use strict-kwargs if you:

  • type-check with ty
  • prefer a standalone linter without plugins
  • want automatic rewrites with strict-kwargs check --fix

Metadata

Release files for strict-kwargs 2026.8.24

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distributions (wheels)

Table of built distributions (wheels) for strict-kwargs 2026.8.24
File Interpreter ABI Platform
strict_kwargs-2026.8.24-py3-none-win_amd64.whl Python 3 none Windows x86-64 Details
strict_kwargs-2026.8.24-py3-none-manylinux_2_39_x86_64.whl Python 3 none Linux glibc 2.39+ x86-64 Details
strict_kwargs-2026.8.24-py3-none-macosx_11_0_arm64.whl Python 3 none macOS 11.0+ ARM64 Details

Total release size: 9.7 MB

Release files / strict_kwargs-2026.8.24-py3-none-win_amd64.whl

Download URL strict_kwargs-2026.8.24-py3-none-win_amd64.whl
Size 3.1 MB
Tags Python 3 Windows x86-64
SHA-256 checksum
How to use checksums
d5c7d4f425a28546700d528d01977e7275e1e4b695862fa7c0ec3fe3c2c7b731
BLAKE2b-256 checksum
How to use checksums
e178094ea71397cf6812cafa5b769c971f4442e304617560b4c0842dcda4862a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 24, 2026.

Transparency log

Release files / strict_kwargs-2026.8.24-py3-none-manylinux_2_39_x86_64.whl

Download URL strict_kwargs-2026.8.24-py3-none-manylinux_2_39_x86_64.whl
Size 3.4 MB
Tags Linux glibc 2.39+ x86-64 Python 3
SHA-256 checksum
How to use checksums
d39d5859eabe9d5a9e4aee0ca7c720fdcf32ac8cce5f0d02f1463e665b7573a0
BLAKE2b-256 checksum
How to use checksums
f8e4a13a6b3c1e6bb7c9a0b0c4bf4f70efd24857e8ff594e951d82eb2004f83a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 24, 2026.

Transparency log

Release files / strict_kwargs-2026.8.24-py3-none-macosx_11_0_arm64.whl

Download URL strict_kwargs-2026.8.24-py3-none-macosx_11_0_arm64.whl
Size 3.2 MB
Tags Python 3 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
d740684600c422586d917ea8139f2b7358221f0669fab074dfdf9218c5509ec9
BLAKE2b-256 checksum
How to use checksums
a6b147e1034415ac31469d81d4aac7070f0d4123de3b4b77759316fa56ef01e5
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 24, 2026.

Transparency log
Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page