AutoSwarm is a Python tool that stands up multiple AI agents like employees, assigns work to them, and hands work over. The public repository is hkuds/autoswarm and the license is MIT. No version packaged as a release has been published yet, and the repository introduces the project as an agent swarm automation tool.
The README describes AutoSwarm as an organization execution tool that splits complex real-world tasks among role-based AI employees. It offers a command-line interface (CLI) and an office-style interface (Office UI) together, and bundles task assignment, handoff, review, and memory into an organization-level execution flow. The repository has 1861 stars.
Self-Built recruits role-based employees to form an organization, Self-Run coordinates task assignment and handoff along with review, and Self-Grown accumulates lessons drawn from work results as the organization's memory and experience. A Task Mode for a single agent is provided as well. It targets many fields, including AI development, software development, finance, sales, media, e-commerce, and education.
The evidence is still shallow. The research record scores freshness 6.0, technology 7.0, quality 7.0, and momentum 3.3, with only momentum coming out low. The repository was also created on 2026-07-01. Claims of broad industry coverage and self-growth are not proven by the repository materials, so they were left out of the scoring rationale. For now it is better treated as a tool worth checking out for how it structures organizations and passes work along, and its real effectiveness needs separate verification.