AgentOS CLI
AgentOS runs a team of OpenClaw agents: workspaces, scheduled tasks, activity reports and human approvals
- 134 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- Runs OpenClaw workers with access to files, models, channels and browser accounts that act on their own on a schedule
- The remote `install.sh` install and a panel exposed on a server widen the access surface, and the administrator password protects it
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/agentosDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Runs OpenClaw workers with access to files, models, channels and browser accounts that act on their own on a schedule
- The remote `install.sh` install and a panel exposed on a server widen the access surface, and the administrator password protects it
This is third-party code. Review the repository files before installing.
What it does
An open-source operating layer for running several agents from one workstation. OpenClaw serves as the agent runtime. In AgentOS you create workspaces and digital workers with `worker`, `setup`, `browser` and `monitoring` presets, give each an identity, context, memory, policies, model and access, and assign one-off, recurring and scheduled tasks. Live activity, transcripts, output files and token usage sit alongside. A worker's context lives in OpenClaw workspace files: `AGENTS.md`, `SOUL.md`, `IDENTITY.md`, `MEMORY.md`. The commands `agentos start --open`, `agentos status` and `agentos doctor --deep` start and check the system. An OpenClaw plugin, `@sapienx/openclaw-agentos`, hands commands to the main CLI, and a Railway template runs a separate Chromium service.
Who it is for. People who run several OpenClaw agents on real projects and want tasks, schedules, costs and approvals in one interface.
Good fit when
- You already run several agents and want one place for tasks, statuses and token spend
- Recurring work should run on a schedule with a report and human intervention
- A new project should become a workspace with a team and first missions
- You need a deployable service with a protected administrator account on Railway
Not a fit when
- You run a single agent in a plain terminal and a control layer is overkill
- Your agent runtime is not OpenClaw: AgentOS is built on its gateway
- You need official compatibility guarantees: the project is independent and fast-moving
Example request
Create an AgentOS workspace for a blog, add an analyst worker and give it a weekly task with a reportLimitations
- The runtime runs on OpenClaw. If it is missing or unhealthy, AgentOS shows a setup or unavailable state and draws no invented data.
- The `curl -fsSL .../install.sh | bash` install runs a remote script, and `npm install -g @sapienx/agentos` is available for the cautious.
- The initial Railway administrator password is set once, and changing the variable later does not change the stored account.
- The OpenClaw plugin only forwards commands to the main CLI and holds no second runtime, so AgentOS is installed separately.
- Sensitive worker actions need your approval, and access policies are configured by hand.
How to disable. Stop the service, remove the package with `npm uninstall -g @sapienx/agentos`, remove the plugin through `openclaw plugins` and delete OpenClaw and AgentOS state from the data directory.
FAQ
What does AgentOS need to run?
The OpenClaw runtime, a configured model and a workspace. If OpenClaw is already installed, AgentOS connects to the live gateway.
How do I install AgentOS?
Run `npm install -g @sapienx/agentos` or `pnpm add -g @sapienx/agentos`, then `agentos start --open`. A `install.sh` script exists for macOS and Linux.
How do I check the installation?
`agentos status` and `agentos doctor --deep` show the state of the environment, gateway and models.
What is the OpenClaw plugin for AgentOS?
A thin plugin from ClawHub: `openclaw agentos`, `status` and `doctor --deep` are forwarded to the canonical AgentOS CLI.
Can I deploy it on a server?
Yes, a Railway template runs AgentOS and the OpenClaw gateway in one service and interactive Chromium in a separate private service, with only the administrator password required.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI agent swarm: a leader spawns workers in separate git worktrees, and they share tasks and exchange messages via files.
The clawctl CLI manages a fleet of AI agents on local machines over SSH and Ansible: install, start, upgrade and token tracking.
An open coding agent for the terminal, browser and desktop that works with models from OpenAI and Claude to Ollama
One runtime for Claude Code, Codex, OpenCode and more: a run counts as done only when checks the agent did not write pass
Self-hosted system for a team of AI agents: org chart, tasks, budgets and oversight of Claude Code, Codex and OpenClaw
An AI-agent workspace with parallel worktrees, cross-provider orchestration, and bots across chat channels.
README in short
The README presents AgentOS as the layer a team of digital workers runs on. A feature tour, a five-minute start, the Railway deploy, the package install and the OpenClaw plugin follow.