iamlukethedev/Herald-OS

An agent-native operating system, with Hermes Agent as the interface. Independent project, not affiliated with Nous Research.

What it solves

Herald OS aims to replace traditional menu-and-folder based operating system interactions with an agent-native interface. It allows users to operate their entire machine—opening apps, organizing files, and running routines—through a conversational AI agent (Hermes) rather than manual navigation.

How it works

It functions as an agent-centric shell or a full operating system (based on Fedora Linux) that integrates the Hermes Agent as its primary interface. It uses a system bridge (herald-os-bridge) to give the AI agent permission-based access to system tools, including process management, file search, and app control. It can run as a standalone OS on a PC, a session on Arch Linux, or a fullscreen application over macOS.

Who it’s for

Users who want an AI-first computing experience where an agent manages the OS and software development tasks, as well as those interested in experimental agentic operating systems.

Highlights

  • Agent-Native Interface: Uses Hermes Agent for system-wide control via text or voice.
  • Studio: A dedicated environment for building software where the agent writes code and provides a live preview in one window.
  • Herald Canvas & Office: AI-integrated image editing and document suites (Docs, Sheets, Slides) that allow the agent to perform edits and generate content.
  • Contextual Overview: Automatically surfaces recent documents, project folders, and calendar events to help users resume work threads.
  • Permission System: Every system-changing action performed by the agent requires user approval through a controlled permission tier.
  • System Integration: Includes tools for crash report explanation, screen content analysis (OCR/QR codes), and automated routines.

Related

  • Project
    blakeandersonw/ai-app-playbook

    A comprehensive guide and technology stack for building, launching, and scaling profitable AI applications while minimizing infrastructure costs.

  • Project
    egeominotti/bunqueue

    A high-performance, infrastructure-light job queue for AI agents and automation that supports embedded SQLite or PostgreSQL backends without requiring Redis.

  • Project
    multikernel/sandlock

    A lightweight, rootless Linux sandbox for isolating untrusted code using Landlock and seccomp, offering high-performance security for AI agents and scripts without the overhead of containers.

  • Project
    tjxj/z-skills

    z-skills is a collection of reusable local Agent Skills covering web scraping, video download, document parsing, audio generation, Markdown-to-Word/PDF, grounded Q&A, and more—designed to plug into Codex or Claude Code for Chinese content creation and automation. Each skill includes trigger words, scripts, and evaluation examples so agents can reliably carry out complex workflows.