Open Arsenal

Progress You Can Audit.

HelmOS is a modular, local-first operating layer designed to wrap and enhance existing AI agents rather than replace them. It introduces a structured governance and execution framework that ensures every agent operates under clear human-defined intent, with explicit verification before actions are executed. Instead of treating agents as autonomous black boxes, HelmOS turns them into governed components within a larger, auditable system.

At its core, HelmOS focuses on aligning intent with execution. Before any meaningful action is taken, the system translates user requests into structured execution plans, validates parameters, evaluates risk, and requires human confirmation when necessary. This intent verification layer is paired with a human-in-the-loop governance model, ensuring that critical operations remain under direct user control while still enabling automation where appropriate.

HelmOS also introduces a Trust Framework that evaluates agents over time based on reliability, accuracy, safety, and compliance. Agents are not assumed to be equally reliable; instead, they earn trust through measurable performance. This trust directly influences permissions, approval requirements, and execution boundaries, creating a dynamic system where behavior determines capability.

To support real-world usage, HelmOS includes a Workflow Marketplace for reusable agent processes, a Memory System for both short-term and long-term context retention, and a comprehensive Evaluation Engine for benchmarking performance and detecting regressions. These systems allow users and organizations to standardize and scale complex workflows without sacrificing transparency or control.

In addition, HelmOS incorporates an AI Constitution Layer that defines immutable governance principles, ensuring that safety, auditability, and human oversight cannot be bypassed or self-modified by agents. Combined with Incident Management, the system continuously monitors failures, reconstructs events, and feeds insights back into improvement cycles.

Together, these features make HelmOS not just an agent framework, but an operating system for governed intelligence—where automation is powerful, but always visible, verifiable, and accountable.

This specification is released under the GNU Affero General Public License v3.0 or later (AGPL-3.0+) and may be used freely with required attribution under Section 7. A Specification Branding License is available for attribution-free deployments, with fees based on usage, scope, and deployment size.