Getting Started¶

Installation¶

Install from PyPI:

pip install miminions

Or with uv:

uv add miminions

For SQLite vector memory support:

pip install miminions[sqlite]

Requirements¶

Python 3.10+
An OpenRouter API key (set as OPENROUTER_API_KEY)

Your First Agent¶

import asyncio
from miminions.agent import create_minion

async def main():
    agent = create_minion("MyAgent")

    def calculator(operation: str, a: int, b: int) -> int:
        if operation == "add":
            return a + b
        elif operation == "multiply":
            return a * b
        return 0

    agent.register_tool("calculator", "Perform arithmetic", calculator)
    reply = await agent.run("What is 6 multiplied by 7?")
    print(reply)

asyncio.run(main())

Using the CLI¶

After install, the miminions CLI is available:

miminions --help

Start an interactive chat session:

miminions chat start

Resume a previous session by ID:

miminions chat start --session <session_id>

Workspace Management¶

# List all workspaces
miminions workspace list

# Create a workspace with sample data
miminions workspace add --name "Demo" --description "Demo workspace" --sample

# Show workspace details
miminions workspace show <workspace_id>

# Evaluate workspace rules
miminions workspace evaluate <workspace_id>

Development Setup¶

git clone https://github.com/MiMinions-ai/MiMinions.git
cd MiMinions
pip install -e ".[dev]"
pytest tests/

Running Tests by Layer¶

# Unit tests
pytest tests/unit/

# Integration tests
pytest tests/integration/

# End-to-end tests
bash tests/e2e/run_all.sh

Publishing¶

Build and publish with uv:

uv build
uv publish

Build a standalone CLI binary:

bash deploy/build_cli.sh