pyproject.toml

[build-system]
requires = ["hatchling"]
build-backend = "hatchling.build"

[project]
name = "judgecheck"
version = "0.2.0"
description = "Validate LLM-as-judge reliability with inter-rater agreement, not accuracy."
readme = "README.md"
requires-python = ">=3.10"
license = "MIT"
license-files = ["LICENSE"]
authors = [{ name = "Alexander Theruviparambil" }]
keywords = ["llm", "eval", "evaluation", "llm-as-judge", "inter-rater-reliability", "kappa"]
classifiers = [
  "Development Status :: 4 - Beta",
  "Intended Audience :: Developers",
  "Programming Language :: Python :: 3.10",
  "Programming Language :: Python :: 3.11",
  "Programming Language :: Python :: 3.12",
  "Programming Language :: Python :: 3.13",
  "Typing :: Typed",
]
# No runtime dependencies. The math is small enough for the standard library,
# and a package that installs anywhere with nothing behind it is worth more
# than one that pulls numpy to do arithmetic this simple.
dependencies = []

[project.optional-dependencies]
dev = ["pytest>=8", "mypy>=1.11", "ruff>=0.6"]
# Third-party implementations used to cross-check the coefficients in
# tests/test_crossvalidation.py. Kept out of `dev` so the common path stays
# dependency-light; CI installs them in a dedicated job.
crossval = ["statsmodels>=0.14", "krippendorff>=0.6", "numpy>=1.24"]

[project.scripts]
judgecheck = "judgecheck.cli:main"

[project.urls]
Homepage = "https://github.com/theruviparambil/judgecheck"
Repository = "https://github.com/theruviparambil/judgecheck"
Issues = "https://github.com/theruviparambil/judgecheck/issues"

[tool.hatch.build.targets.wheel]
packages = ["src/judgecheck"]
# Ships the PEP 561 marker so downstream type checkers actually see the
# annotations. Without it the "Typing :: Typed" classifier is a lie.
artifacts = ["src/judgecheck/py.typed"]

[tool.pytest.ini_options]
testpaths = ["tests"]
addopts = "-q"
markers = [
  "slow: spawns subprocesses (hash-seed determinism checks)",
]

[tool.mypy]
# 3.12, not our 3.10 floor: the cross-validation extras ship stubs using
# 3.12-only syntax that mypy cannot even parse under an older target. Real
# 3.10 compatibility is enforced by running the full suite on 3.10 in CI,
# which catches runtime breakage that a type check would not.
python_version = "3.12"
strict = true
# scripts/ is checked too: mutation_sweep.py is imported by the README-claims
# test, and a verification tool that is not itself verified is worth little.
files = ["src/judgecheck", "tests", "scripts"]
mypy_path = "scripts"

[tool.ruff]
line-length = 100
target-version = "py310"

[tool.ruff.lint]
select = ["E", "F", "I", "UP", "B", "SIM", "RUF"]

# We do not type check other people's libraries.
[[tool.mypy.overrides]]
module = ["numpy.*", "statsmodels.*", "krippendorff.*", "scipy.*", "pandas.*"]
follow_imports = "skip"
ignore_missing_imports = true