sleazy

A tiny utility to build CLI tools from Python TypedDict schemas.
Define your parameters with type hints (and Annotated for positionals), then call parse() to get a typed dict and stringify() to turn it back into args.

Installation

uv pip install sleazy

Example CLI

from typing import Annotated, Literal, TypedDict
from sleazy import parse, stringify


class AppConfig(TypedDict):
    # 1) Required positional (exactly one string)
    input_file: Annotated[str, 1]

    # 2) Optional positional (zero or more strings)
    tags: Annotated[str, "*"]

    # 3) Optional keyword args
    output: str  # string
    verbose: bool  # flag
    retries: int  # integer
    mode: Literal["auto", "manual"]  # choice via Literal or enum


if __name__ == "__main__":
    # parse() returns a dict matching AppConfig
    args = parse(AppConfig) # python example.py data.txt tag1 tag2 --output out.txt --verbose --retries 5 --mode auto
    print(args)
    # {
    #   'input_file': 'data.txt',
    #   'tags': ['tag1', 'tag2'],
    #   'output': 'out.txt',
    #   'verbose': True,
    #   'retries': 5,
    #   'mode': 'auto'
    # }

    # stringify back to CLI args, using AppConfig to keep positionals first
    print(stringify(args, AppConfig))
    # [
    #   'data.txt',
    #   'tag1', 'tag2',
    #   '--output', 'out.txt',
    #   '--verbose',
    #   '--retries', '5',
    #   '--mode', 'auto'
    # ]

    # Without the TypedDict, stringify treats everything as flags
    print(stringify(args))
    # [
    #   '--input-file', 'data.txt',
    #   '--tags', 'tag1', '--tags', 'tag2',
    #   '--output', 'out.txt',
    #   '--verbose',
    #   '--retries', '5',
    #   '--mode', 'auto'
    # ]

Notes

• Positionals are marked via Annotated[type, count] (1, ?, *, +, or exact integer).
• All other fields become --field-name flags by default.
• It's useful to pass your TypedDict type to stringify() so it knows which fields are positional (and their order). Without it, everything is emitted as a keyword flag.