nanocli 0.1.0

Recursive hierarchical CLI framework with typed configs and YAML synthesis

NanoCLI

A cli framework with the simplicity of argparse, the colors of rich, the config handling of Hydra, and without the complexity.

Core Model

CLI = YAML Tree
├── commands (leaf nodes)
└── groups (subtrees)

Quick Start

from dataclasses import dataclass
from nanocli import group

@dataclass
class TrainConfig:
    epochs: int = 100
    lr: float = 0.001

app = group()

@app.command()
def train(cfg: TrainConfig):
    print(f"Training for {cfg.epochs} epochs")

if __name__ == "__main__":
    app()

Usage

# Run command
python app.py train
python app.py train epochs=200

# Hydra-style overrides at root level
python app.py train.epochs=200 -p

# Print config: -p (local), -g (global from root)
python app.py train -p
python app.py data download -g

# Load YAML config
python app.py -c config.yml train

# Help
python app.py -h
python app.py train -h

Nested Groups

app = group()

@app.command()
def train(cfg: TrainConfig):
    ...

data = app.group("data", help="Data commands")

@data.command()
def download(cfg: DownloadConfig):
    ...

python app.py data download
python app.py data download dataset=coco -p
python app.py data download dataset=coco -g  # prints full tree

Flags

Flag	Meaning
-p	Print config from current node
-g	Print config from root (global)
-h	Show help
-c PATH	Load base config from YAML

Development

make dev          # Install with dev deps
make test         # Run tests
make pre-commit   # Run all checks
make lint-fix     # Fix lint issues
make type-check   # Type check

License

MIT