agent0ai/agent-zero

Agent Zero AI framework

Agent Zero – an open‑source AI‑agent workbench

What it is – Agent Zero is a Docker‑based framework that gives a language‑model‑driven agent a full Linux desktop to work in. The agent can drive GUI applications, a built‑in browser, LibreOffice editors, and a terminal, while you watch and intervene through a web UI called the Canvas.

Key components

  • Linux desktop (XFCE) inside the container – the agent can open any GUI program (e.g., Blender, VS Code, terminal) and manipulate files just like a human user.
  • Browser with DOM annotation – click elements to inspect, modify, “lift” (capture) components, or leave comments that the agent can act on.
  • Live document coworking – Markdown, ODT/ODS/ODP editors are shared between you and the agent in real time.
  • Plugin Hub – >100 community plugins for skills, memory back‑ends, development frameworks, UI tweaks, etc.
  • Projects & memory – isolated workspaces that keep files, secrets, Git repos, and model presets separate per task.
  • A0 CLI connector – a thin client that bridges the Dockerised agent to your host machine, letting the same agent edit local repositories or run host‑side tools.
  • Multi‑agent cooperation – an agent can spawn sub‑agents to specialize (research, coding, analysis) and aggregate their results.
  • Time‑Travel snapshots – automatic versioning of the /a0/usr workspace for easy diff/rollback (supplementary to Git).

Installation

  • A0 Launcher (GUI) – download a platform‑specific binary (macOS, Linux, Windows) that handles Docker checks, instance creation, and port management.
  • A0 Install (scripted) – one‑line curl/PowerShell command that sets up a Docker container and optionally runs in quick‑start mode.
  • Direct Docker run – docker run -p 80:80 -v a0_usr:/a0/usr agent0ai/agent-zero for manual control.

Typical workflow

  1. Start an instance via the Launcher or script.
  2. Open the web UI, configure an LLM provider (OpenAI Codex, Gemini, Claude, etc.).
  3. Create a Project (optionally clone a Git repo). The agent now has its own isolated filesystem.
  4. Ask the agent to perform a task – e.g., “design a 3‑D logo in Blender”, “annotate this web page and re‑implement the hero component in React”, or “build a budget spreadsheet”.
  5. Watch the agent act in the Canvas desktop or browser; intervene by typing, clicking, or editing files directly.
  6. Use plugins or custom skills to extend functionality, or spawn sub‑agents for larger jobs.
  7. When finished, snapshot the workspace or revert via Time Travel.

Use‑case highlights

  • Software engineering – the agent can open a codebase, run tests, edit files, and commit changes while you supervise.
  • Design & UI iteration – browse competitor sites, annotate elements, and have the agent generate matching components.
  • Financial analysis – collect data, build ODS spreadsheets, and produce editable charts.
  • Document production – co‑author Markdown, ODT, or presentation files without leaving the chat.
  • Automation & orchestration – schedule recurring checks, run CLI tools on the host via the A0 connector, or chain multiple specialized sub‑agents.

Safety notes

  • Run inside Docker (or another container) to isolate the agent.
  • Limit host‑directory mounts; grant read/write only where needed.
  • Store credentials in project secrets, not in prompts.
  • Review any actions that affect production systems, money, or private data.

Where to learn more

  • Full docs: ./docs/ (quick‑start, browser guide, desktop guide, plugins, memory, MCP, etc.)
  • Community: Discord, YouTube channel, Skool community, and GitHub Issues.
  • Open‑source code and plugin hub are on GitHub under the agent0ai organization.

Bottom line – Agent Zero is a comprehensive, extensible platform that lets LLM agents operate in a real Linux desktop environment, bridging the gap between pure text‑only chat and hands‑on software development, design, and data work.

Related

  • Project
    Twigpine/zero

    Zero is a Go‑based terminal AI coding assistant that lets you talk to LLMs, have them edit files, run commands, and manage persistent sessions, with fine‑grained permission controls and support for many model providers.

  • Dispatch
    Docker Agent v1.149.0 – Open‑Source AI Agent Builder and Runtime

    Docker Agent v1.149.0 is an open‑source CLI plugin that lets you declaratively define, run, and share multi‑agent AI workflows using YAML, with support for many model providers and a rich tool ecosystem.

  • Project
    uvwt/agentdock

    AgentDock is an open‑source runtime that lets LLM chat agents (e.g., ChatGPT) safely perform file operations, run commands, use Git, automate browsers, and coordinate work across multiple computers or containers via a secure HTTP‑based MCP endpoint. It provides fine‑grained permission controls, task persistence, and extensible Skills, but does not include its own chat UI or model inference.

  • Project
    Agenta-AI/agenta

    Agenta is an open‑source, self‑hostable workspace for building AI agents that can be chatted with, shared with a team, and run in the background. It supports many LLM providers, integrates with thousands of SaaS tools, offers permission controls, tracing, and scheduling, and provides a web UI plus a rich developer ecosystem.

  • Project
    agentconnect-md/agentconnect

    AgentConnect is an open‑source, self‑hosted platform that lets teams run many AI agents (Claude Code, OpenAI Codex, Grok Build, etc.) inside Slack, Discord, Telegram, Google Chat, GitHub, GitLab, Linear and other tools. Agents can call each other, keep memory, be routed by Jev, and act on webhooks or schedules. The system consists of Daemons (run agents), an optional Relay, and a Control‑Plane UI for config, auth and observability. It ships via Docker‑Compose for local use and a Helm chart for Kubernetes, all under Apache‑2.0.