Workspace-based operations Mission control for agents Live execution visibility Human approvals and oversight OpenClaw-powered foundation
Repo
AgentOS by SapienX
Watch demo Quickstart

The human layer for OpenClaw

One human. Thousands of agents.
Hundreds of projects.

AgentOS is the control layer that simplifies complex agent, project, and AI workforce management on top of OpenClaw.

OpenClaw integration catalog

Works With Everything

AgentOS works seamlessly across OpenClaw integrations, bringing every connection into one operating layer for execution. AgentOS unifies OpenClaw integrations into one operating layer for execution.

View all integrations
Operator Core

A Smarter Way to Operate OpenClaw

OpenClaw provides the runtime. AgentOS makes it legible, guided, and operable for real people doing real work.

01 Fast start

Guided Onboarding

Get from zero to a working system faster with a cleaner setup flow on top of OpenClaw. Start faster with a simpler setup flow.

Cleaner setup Faster first success
02 Guided build

Workspace, Agent & Task Wizards

Create better workspaces, agents, and tasks through guided flows powered by the full flexibility of OpenClaw. Create workspaces, agents, and tasks with guided flows.

Guided creation Flexible underneath
03 Human layer

Human Control Surface

Run agents, missions, and operations from one clear interface built for human operators. Run agents and operations from one clear interface.

One interface Operational clarity
04 Live signal

Runtime Visibility

See what agents are doing, what changed, and what needs your attention in real time. See agent activity and important changes in real time.

Live changes Attention cues
05 Review gates

Approval Layer

Review critical actions before execution and keep humans in control when it matters most. Review critical actions before they run.

Critical checkpoints Human-in-loop
06 Focused teams

Job & Team Builder

Create custom jobs and purpose-built agent teams for focused operational work like Telegram, growth, or research. Build custom jobs and focused agent teams.

Custom jobs Purpose-built teams
QuickStart

Install AgentOS.

Pick an install path, run one command, and verify the local runtime.

CLI beta

# Fastest path. The one-liner auto-detects your platform and installs the right AgentOS release.

$ 
curl -fsSL https://raw.githubusercontent.com/SapienXai/AgentOS/main/install.sh | bash
Auto-detecting platform Launch with agentos start --open Verify with agentos doctor

Start with the one-liner unless your setup already depends on pnpm or a source checkout.

Operating Stack

Built above

AgentOS turns OpenClaw into a human-operable control system for real work.

01

Live runtime

See sessions, models, transcripts, presence, and gateway state as they actually are.

Sessions Models Transcripts Presence Gateway state
02

Real workspace

Work directly on real files, folders, docs, memory, and deliverables.

Files Folders Docs Memory Deliverables
03

Human control

Plan, inspect, approve, and steer agents without losing the full operating picture.

Plan Inspect Approve Steer Operating picture
Foundation OpenClaw

Runtime, workspace state, and control loops stay grounded in the system underneath.

Start with AgentOS

You are not just using AI. You are operating AI labor.

Run agents, projects, and operations from one control surface designed for real operators. Built on OpenClaw, AgentOS gives you structure, visibility, and control across workspaces, tasks, and execution.

Open repository Watch demo