artificer-cli 0.1.0a4

Shared CLI interface for Artificer — AI dev tooling built on MCP

artificer-cli

The shared CLI interface for the Artificer project — AI dev tooling built on MCP (Model Context Protocol).

What is this?

A minimal, configuration-driven CLI that dynamically loads features from installed Artificer packages. Instead of shipping a monolithic CLI, each Artificer package (workflows, conversations, etc.) registers its own commands through a simple plugin system.

Installation

pip install artificer-cli

Usage

artificer --help
artificer --version

Commands are added by installing Artificer feature packages and registering them in your pyproject.toml:

[tool.artificer]
features = [
    "artificer.workflows.feature.WorkflowFeature",
    "artificer.conversations.feature.ConversationFeature"
]

Creating Features

Features extend ArtificerFeature and implement register():

from artificer.cli.feature import ArtificerFeature
from artificer.cli.config import ArtificerConfig
import click

class MyFeature(ArtificerFeature):
    @classmethod
    def register(cls, cli: click.Group, config: ArtificerConfig) -> None:
        @cli.command()
        def my_command():
            """Does something useful."""
            click.echo("Hello from my feature")

Architecture

• config.py — Loads [tool.artificer] from pyproject.toml
• feature.py — Base class for all features
• loader.py — Dynamic feature import and validation
• cli.py — Builds the Click CLI from config

The artificer/ directory is a PEP 420 namespace package, allowing other packages to extend it (e.g., artificer.workflows, artificer.mcp).

Development

# Install dev dependencies
uv sync

# Run tests
./scripts/test.sh

# Type check
./scripts/typecheck.sh

# All checks
./scripts/check.sh

License

MIT