Dashboard Skill

Build and run a Kanban-style task management dashboard for your Clawdbot agent.

Overview

This skill creates a local web dashboard that lets you:

• Manage tasks in a Kanban board (To Do → In Progress → Done → Archived)


• Monitor agent status in real-time (Online/Thinking/Ready)


• Drop notes for the agent to check on heartbeats


• View action logs and quick-access deliverables

Quick Start

Run the setup script to install and start the dashboard:

bash /path/to/skills/dashboard/setup.sh

Or have your agent run it:

Set up and start my Clawdbot dashboard

Manual Setup

1. Install Dependencies

# Ubuntu/Debian
apt-get install -y python3-flask python3-flask-cors

# Or with pip
pip3 install flask flask-cors

2. Create Dashboard Directory

mkdir -p ~/clawd-dashboard/{templates,static/css,static/js,data}

3. Copy Files

Copy these files from the skill's src/ directory:

• app.py → ~/clawd-dashboard/


• templates/index.html → ~/clawd-dashboard/templates/


• static/css/style.css → ~/clawd-dashboard/static/css/


• static/js/dashboard.js → ~/clawd-dashboard/static/js/

4. Start the Dashboard

cd ~/clawd-dashboard
python3 app.py

Dashboard runs on **

Configuration

Environment Variables

Variable

Default

Description

CLAWD_WORKSPACE	/root/clawd	Path to agent workspace
DASHBOARD_PORT	5050	Port to run dashboard on

Customization

Edit app.py to customize:

• Deliverables: Modify get_deliverables() to add your own folder shortcuts


• Agent ID: Change the session key pattern in get_agent_status() if your agent isn't named "clawd"

Features

Kanban Board

• Drag & drop tasks between columns
• Color-coded columns (red=todo, yellow=progress, green=done, gray=archived)
• Click to edit any task
• Tasks persist in data/tasks.json

Agent Status

• Polls clawdbot status --json every 10 seconds
• Shows: Online/Offline, Thinking/Ready state
• Displays model name and token usage

Notes Section

• Add instructions for your agent
• Agent can read these during heartbeat checks
• Saved to data/notes.json

Action Log

• Tracks task creates/moves/deletes
• Timestamps all actions
• Last 100 entries stored in data/action_log.json

Running as a Service

To keep the dashboard running after logout:

# Using systemd (recommended)
cat > /etc/systemd/system/clawd-dashboard.service << 'EOF'
[Unit]
Description=Clawd Dashboard
After=network.target

[Service]
Type=simple
WorkingDirectory=/root/clawd-dashboard
ExecStart=/usr/bin/python3 app.py
Restart=always
Environment=CLAWD_WORKSPACE=/root/clawd

[Install]
WantedBy=multi-user.target
EOF

systemctl daemon-reload
systemctl enable clawd-dashboard
systemctl start clawd-dashboard

Security Notes

• Dashboard has no authentication by default
• Only run on trusted networks (localhost, Tailscale, VPN)
• Do NOT expose to public internet without adding auth

Troubleshooting

"Status shows offline"

• Check that clawdbot status --json works from command line
• Verify the agent ID pattern matches your setup

"Tasks not saving"

• Check that data/ directory exists and is writable
• Look for errors in the terminal or /tmp/dashboard.log

"Port already in use"

• Change port: python3 app.py then edit the app.run() line
• Or kill existing process: pkill -f "python3 app.py"

Files Reference

clawd-dashboard/
├── app.py                 # Flask backend (API + routes)
├── data/
│   ├── tasks.json         # Kanban tasks
│   ├── notes.json         # Notes for agent
│   └── action_log.json    # Activity history
├── static/
│   ├── css/
│   │   └── style.css      # Dark theme styles
│   └── js/
│       └── dashboard.js   # Frontend logic
└── templates/
    └── index.html         # Main page template

Credits

Inspired by the "Klaus Dashboard" from Nate Herk's Clawdbot video.