Antigenic operating system for my AI agents Claude Code, open claw, codex

The developer project Dodaios launched an open-source agent operating system designed to run coding agents, web browsers, and desktop software locally through voice commands or text chat. According to github.com's report, the platform operates entirely on the user's hardware across Windows, macOS, Linux, and headless server environments without requiring an account, paid subscription tier, or intermediary cloud service. Developers connect their own credentials, utilizing existing OpenAI, Claude, or ChatGPT plans, hosted API endpoints from Gemini, OpenRouter, and NVIDIA, or local inference models served through Ollama and OpenAI-compatible backends.

This report examines the architecture of Dodaios, formerly known as Personal Jarvis, detailing how its interface orchestrates multi-agent coding sessions, local speech recognition, and system access controls. The analysis is structured for software engineers, DevOps practitioners, and systems developers who need to evaluate how local agent orchestration operates alongside command-line developer tools, how the software handles execution privileges, and how teams can deploy the system across desktop and remote workstation environments.

What shipped in Antigenic operating system for my AI agents

The release introduces the Agentic IDE, a unified workspace that splits a single monitor into concurrent terminal panes where different coding assistants operate simultaneously. Each active pane receives an individual call sign such as T1 or T2, turning the window into an interactive terminal cluster. Developers can run Claude Code, Codex, OpenCode, Kimi Code, GLM, Grok Build, Antigravity, Cursor CLI, and DeepSeek Harness side by side. Next to the execution panes, the interface embeds an integrated code editor, visual Git controls, diff views, and pull request management screens so engineers inspect file modifications directly before applying patches.

Beyond terminal panes, Dodaios provides specialized persistent agents for research, release management, and technical documentation. Each autonomous specialist maintains a permanent conversation thread, learns incrementally across completed tasks, executes scheduled routines, and writes documents, reports, and dashboards to a shared display wall called Artifacts. Specialized agents also run across runtime engines such as Hermes and OpenClaw. System-level activities integrate a local wiki engine called Memory that agents search, update, and cite with knowledge maps, alongside a spatial ship interface named Verse that shows on-duty agents.

What improved in Antigenic operating system for my AI agents

Workflow coordination advances beyond standard single-terminal command-line interfaces through an integrated voice engine and a tiered safety model. Offline speech recognition handles natural speech, custom wake phrases, and dictation without sending audio streams to remote servers. When an operator issues verbal instructions, the software converts sentences into repository-aware briefs, routing commands to designated panes like instructing T1 to run unit tests and T2 to resolve failures. If a designated call sign is ambiguous, the operating system pauses to request clarification before assigning tasks to background processes.

Safety governance improves by routing every tool execution through a structured risk-evaluation filter categorized into safe, monitor, ask, and block stages. The operating system intercepts any operation capable of modifying the local filesystem or altering network states, requiring explicit user approval before execution proceeds. Every interaction, command run, and tool invocation is logged into an auditable runtime record, preventing autonomous scripts from making unmonitored alterations to operating files or development repositories.

What you gain from Antigenic operating system for my AI agents

Deploying the agent operating system centralizes software automation without surrendering control of source code or paying platform subscription premiums. Organizations avoid vendor lock-in by routing work across local models through Ollama alongside commercial endpoints from OpenAI, Claude, and Gemini without creating third-party accounts. Because all repository data, system logs, and agent memories remain on the workstation, sensitive intellectual property stays behind local firewalls while remaining accessible to automation routines that build custom skills and automations from repeated user tasks.

Engineering teams also gain access to an extensible integration layer that incorporates more than 40 preconfigured plugins, arbitrary Model Context Protocol servers, command-line executables, and custom operational skills. The built-in marketplace allows users to install community-built agents, routines, and plugins with a single interaction. Desktop and browser computer use capabilities enable agents to navigate graphical applications, while automated background routines run recurring integration checks, generate technical reports, and maintain project documentation without stalling interactive developer terminals.

How to get Antigenic operating system for my AI agents

Dodaios provides automated installation scripts for desktop operating systems that download private runtimes for Python and Git without requiring administrator privileges. On Windows workstations, users install the software through PowerShell:

irm https://raw.githubusercontent.com/Dodaios/Dodaios/main/install/install.ps1 | iex

For macOS and Linux systems, the setup executes through a single shell command:

curl -fsSL https://raw.githubusercontent.com/Dodaios/Dodaios/main/install/install.sh | bash

Rerunning the installation scripts checks for updates and refreshes the installation files. After setup completes, the desktop application launches from the system menu or by entering the desktop launch command in a terminal:

dodaios

Developers managing remote or headless machines can install the package through Python's package manager and host the web interface across port 47821:

pip install dodaios
dodaios serve

Precompiled packages are distributed as Windows x64 binaries, Linux x86_64 AppImages, and macOS disk images for both Apple Silicon and Intel hardware.

What to watch after Antigenic operating system for my AI agents

Operating system integration requires attention to security certificates and platform permissions during initial rollout. The macOS disk image is not currently notarized, requiring administrators to manually approve execution under System Settings Privacy and Security by clicking Open Anyway. On Windows, executable signing is transitioning to the SignPath Foundation program with validation certificates supplied by SignPath.io. Users deploying headless instances must also configure HTTPS bindings to enable remote microphone permissions for web-based speech recognition.

Long-term project governance reflects an evolving licensing structure that users must track across versions. While releases up to and including version 1.6.0 remain governed by their original MIT license, subsequent releases distribute under the Apache 2.0 license. In addition, pip packaging across Intel macOS hardware and Windows ARM64 platforms requires pointing package managers to native package indexes. Development contributions remain organized through GitHub issues, documentation tickets, and community discussions under the project's security and trademark guidelines.