minions

Async-native orchestration framework for running many long-lived “minions” (workers) on a single machine, with persistence and structured workflows.

Status: Pre-alpha (0.0.x). APIs are experimental and may change without notice.

What is this?

minions is an opinionated orchestration framework for:

• running multiple long-lived async workers (“minions”)
• wiring them into structured workflows / pipelines
• managing shared resources with explicit lifecycles
• persisting state (like via SQLite/aiosqlite) so workflows can resume after restarts

It’s aimed at things like:

• trading / bot fleets
• data collection loops
• background automation agents
• single-node orchestrated “systems” that don’t warrant full Kubernetes-style infra

The core ideas:

• Pipelines – long-lived event emitters that produce work items
• Minions – long-running async workers that listen to pipelines and spawn workflows
• Workflows – per-event units of work that run to completion using shared resources
• Resources – shared dependencies with startup/shutdown semantics
• Persistence – minimal, pragmatic state so the system can recover and resume

Project status

This project is under active development and still in the pre-0.1.0 design/implementation phase:

• APIs and naming are still being refined
• Tests and docs are evolving
• Breaking changes are expected between 0.0.x versions

If you’re here early, treat the codebase as a workbench, not a stable library.

Installation

The package name is reserved on PyPI, but the project is not ready for general use.

If you still want to experiment at your own risk:

pip install minions