Give your virtual office a name. This appears in the sidebar and page title.
2. CONNECT TO OPENCLAW
Tell the Virtual Office where your OpenClaw installation lives so it can find your agents.
This is the folder that contains your openclaw.json and agents/ directory.
📁 Where to find it:
Linux/Mac (most common): /home/youruser/.openclaw
Running Virtual Office in Docker?
Use the path you mounted with -v. For example: docker run -v /home/eli/.openclaw:/openclaw ...
→ enter /openclaw
Not sure? Run this in your terminal: ls ~/.openclaw/openclaw.json
If it prints a path, use the folder part (e.g. /home/youruser/.openclaw).
🔌 Gateway Connection
Configure the gateway to allow this browser to connect for live agent status and chat.
Auth token for chat & live status. Auto-read from OpenClaw config. Override to connect to a different instance.
🔧 No agents found? Here's what to check:
1. Is the path correct?
The folder should contain openclaw.json and an agents/ directory.
2. Do you have agents configured?
Open your openclaw.json and look for "agents" → "list". Each agent needs a workspace directory.
3. Running in Docker?
Make sure you mounted your OpenClaw directory into the container: docker run -v /path/to/.openclaw:/openclaw ...
4. Permissions?
The Virtual Office server needs read access to the agents/ folder.
Still stuck? You can skip this step and set the path later from ☰ Menu. The office will work — agents just won't appear until connected.
3. AGENT BROWSER (OPTIONAL)
Connect a browser so your agents can browse the web and you can watch live. You can skip this and set it up later.
The address where Chrome/Chromium remote debugging is running.
URL for the live browser view (KasmVNC, noVNC, etc.). This is what shows in the floating browser window so you can watch your agents browse in real time.
What is this? The Agent Browser gives your AI agents a real web browser you can watch live. You need two things:
1. CDP (Chrome DevTools Protocol) — how agents control the browser 2. VNC Viewer — how you watch the browser live in the office
Easiest setup — one Docker command: docker run -d --name vo-browser \ -p 9222:9222 -p 6902:6902 -p 6901:6901 \ kasmweb/chromium:1.17.0
This gives you both:
• CDP URL: http://127.0.0.1:9222
• Viewer URL: http://127.0.0.1:6902 (HTTP, no cert needed)
Or use your own Chrome + VNC: google-chrome --remote-debugging-port=9222 --user-data-dir=/tmp/vo-browser
Then point Viewer URL to your VNC/noVNC server.
Not sure? Skip this step — everything else works without it. You can set it up later from ☰ Menu.
4. SMS PANEL (OPTIONAL)
Let an AI agent send and receive SMS messages through your office. You can skip this and set it up later.
This agent will be shown as the sender/receiver in the SMS panel.
All three Twilio fields are required for sending SMS.
What is this? The SMS Panel lets you send and receive text messages through an AI agent in your office.
What you need: 1. A Twilio account (free trial works) 2. A Twilio phone number (you get one free with trial) 3. Your Account SID and Auth Token from the Twilio Console dashboard
How to get started:
• Sign up at twilio.com/try-twilio
• Copy your Account SID and Auth Token from the dashboard
• Get a phone number from Console → Phone Numbers → Buy a Number
• Paste all three values above
Not sure? Skip this step — everything else works without it. You can enable it later from ☰ Menu.
5. PC PERFORMANCE (OPTIONAL)
Monitor a PC's CPU, RAM, and GPU stats live in your office dashboard. You can skip this and set it up later.
The address where the PC Metrics Server is running. This can be a local or remote machine.
What is this? The PC Performance panel shows live CPU, RAM, and GPU usage graphs in your office sidebar.
How it works:
You run a small Python script on the PC you want to monitor. It exposes metrics on a port, and the Virtual Office reads from it.
Quick setup (on the target PC): pip install psutil python pc-metrics-server.py 8099
The script is included with the Virtual Office at /app/pc-metrics-server.py. Copy it to any PC you want to monitor.
For GPU monitoring (NVIDIA only): pip install pynvml
URL examples:
• Same machine: http://127.0.0.1:8099
• LAN PC: http://192.168.1.100:8099
• Tailscale: http://100.x.x.x:8099
Not sure? Skip this step — you can enable it later from ☰ Menu → Settings.
6. YOU'RE ALL SET!
Your Virtual Office is ready. You can always change settings later from the ☰ Menu button in the toolbar.
What you can do: • ✏️ Edit Office — Place furniture, resize canvas, paint walls & floors
• 👤 Agents — Customize agent appearance, assign desks, set branches
• 💬 Chat — Talk to any agent through the chat panel
• ☰ Menu — Change office name, connection settings, export/import config
Want accurate presence? Install the optional Virtual Office skill so agents report their working/idle status. See /setup docs for details.
☁️ API Usage: The sidebar automatically detects your AI providers (Anthropic, OpenAI, Google, etc.) from your OpenClaw config and shows live usage, quotas, and API key health. No extra setup needed — just connect to OpenClaw above.