Intelligent Software Corp. (Headquarters: Chuo-ku, Tokyo; CEO: Akihiro Hasegawa) is pleased to announce that we have released OpenSpoke, our Kubernetes multi-cluster management platform, as open source software. OpenSpoke is available on GitHub under the Apache License 2.0, and anyone is free to use, modify, and redistribute it.

Repository: https://github.com/OpenSpoke/openspoke

OpenSpoke is designed as the foundation on which AI agents powered by large language models (LLMs) can autonomously carry out the entire software lifecycle — development, verification, deployment, and operations. Human involvement is deliberately narrowed to approving high-risk actions the AI proposes, while day-to-day development and operations cycles run in full automation.

Background

Generative AI has dramatically automated code generation itself. However, the downstream work of verifying that generated code, deploying it safely across multiple clusters, and continuously monitoring and repairing it in production still depends heavily on human hands.

Through our own work operating service infrastructure across on-premises and cloud Kubernetes clusters, we designed and ran a GitOps foundation with structural guardrails specifically built so that AI agents can operate the infrastructure without misfiring.

Key Features

1. End-to-end automation of development, verification, deployment, and operations by AI agents

The central hub and each spoke are wired together through MCP (Model Context Protocol) servers, giving AI agents a unified interface to autonomously perform code edits, PR creation, builds, delivery to multiple clusters, live monitoring, and incident response.

2. Complete auditability through GitOps

All cluster state changes flow through a Git repository as the single source of truth; direct operations such as kubectl apply are prohibited in principle.

3. Structural prevention of AI runaways via operational ledger and policy layer

OpenSpoke ships with a policy layer that declaratively manages, per module, whether it can be restarted, whether it can be edited, and whether the AI is allowed to act alone or must obtain human approval.

4. Hub-and-spoke multi-cluster management

From a single central hub cluster, OpenSpoke uniformly manages downstream clusters — on-premises (RKE2 and others), AWS EKS, Linode, and more.

5. Lightweight architecture built on standalone Fleet

OpenSpoke is built on standalone Fleet — Fleet without Rancher — so it requires no additional control plane or commercial license.

6. Dashboard for AI-operation visibility

A single web dashboard provides a bird's-eye view of every cluster, every module (Bundle), and every action taken by the AI agents.

Use Cases

• An AI-Driven DevOps foundation where AI agents autonomously go from code changes to production deployment.
• The substrate that lets a small team hand a large multi-cluster environment over to AI operations.
• A research and implementation platform for LLM-based infrastructure automation (AIOps).
• A unified operational foundation for hybrid Kubernetes environments that span on-premises and cloud.

Roadmap

We will continue to maintain OpenSpoke and to incorporate feedback from the community. OpenSpoke is also being developed as the core foundation of our Continuous Spawning framework.

Company Profile

Press Contact

Intelligent Software Corp.
Public Relations
Email: [email protected]