agent0ai/agent-zero

Agent Zero AI framework

What it solves

Agent Zero is an open agent framework designed for complex tasks that go beyond simple chat. It provides AI agents with a fully functional, isolated Linux desktop environment, allowing them to interact with real GUI software, manage files, and use a browser with advanced interaction capabilities, rather than being limited to text-based API calls.

How it works

The framework runs within a Dockerized Linux environment (XFCE desktop). It integrates a native browser with "Annotate mode," which allows users to click page elements to give the agent specific instructions (like "change this button's color" or "lift" a component for re-implementation). It also features a live cowork system where users and agents can edit Markdown files and LibreOffice documents (Writer, Calc, Impress) simultaneously.

To extend its capabilities, it uses a Plugin Hub with over 100 community plugins, supports MCP (Model Context Protocol), and provides an A0 CLI Connector to bridge the agent's actions to the agent's host machine's local repositories and shell.

Who it’s for

It is built for software engineers, data analysts, and designers who need an AI agent that can drive desktop applications, perform web-based QA, conduct financial analysis using spreadsheets, and work directly within local codebases.

Highlights

  • Full Linux Desktop: Agents can run real GUI software (e.g., Blender) and use terminals inside the Canvas.
  • Browser DOM Annotation: Interactive surface to inspect, change, or capture web elements for the agent to implement.
  • Live Coworking: Real-time shared editing of Markdown and LibreOffice documents.
  • A0 CLI Connector: Grants the agent controlled access to the host machine's files and shell.
  • Multi-Agent Cooperation: Ability to create subordinate agents for specialized research, coding, or analysis tasks.
  • Time Travel: Snapshot history and diff inspection for recoverable workspace changes.
  • Plugin Hub: Access to 100+ community plugins for memory systems, tools, and workflow orchestration.

Related

  • Project
    agent0ai/dox

    A lightweight framework using a hierarchy of AGENTS.md files to give AI agents precise project context and prevent blind edits during coding.

  • Project
    browser-use/browser-use

    A library and framework that enables AI agents to control a web browser like a human to automate tasks such as filling forms and extracting data.

  • Project
    qqqqqf-q/Arkloop

    An open-source AI agent platform featuring a zero-config desktop app with sandboxed code execution, multi-model routing, and persistent memory.

  • Project
    Gitlawb/zero

    A terminal-based AI coding agent that lets you choose your own LLM provider and maintain strict local control over file edits and shell commands.

  • Project
    Intelligent-Internet/ii-agent

    An open-source AI agent for professional work that enables users to build apps, conduct deep research, and automate workflows using their own LLM API keys.