功能描述

A modern web-based dashboard for managing OpenClaw agents with real-time monitoring, token usage tracking, skill management, and multi-language support. Prov...

使用说明 (SKILL.md)

Agent Dashboard V2

Name: digital staff
Author: mikezhuyan

A modern, feature-rich web dashboard for OpenClaw agent management with real-time monitoring, token usage tracking, and comprehensive agent operations.

Features

📊 Real-time Monitoring: Monitor agent status, token usage, and session statistics
🎨 Modern UI: Dark theme with beautiful animations and responsive design
🛠️ Skill Management: Enable/disable skills for each agent with blacklist support
🔀 Subagent Dispatch: Visual management of agent delegation and permissions
🤖 Agent Creation: Create new agents with guided configuration
🌐 Multi-language: Automatic language detection (English/Chinese)
🌤️ Weather Widget: Real-time local weather and time display
📱 Responsive Design: Works on desktop and mobile devices

Installation

Quick Install

npx skills add mikezhuyan/agent-dashboard@agent-dashboard-v2

Manual Installation

# Clone the repository
git clone https://github.com/mikezhuyan/agent-dashboard.git ~/.agents/skills/agent-dashboard-v2

# Navigate to the directory
cd ~/.agents/skills/agent-dashboard-v2

# Install (optional - creates systemd service)
./install.sh

Usage

Starting the Dashboard

# Run directly
python3 dashboard_server.py

# Or use the installed service
systemctl --user start agent-dashboard

Access

Open your browser and navigate to:

http://localhost:5181

Dashboard Features

1. Agent Grid View

View all agents with status indicators
Drag to reorder agent cards
Click cards to view details in sidebar
Multiple view modes (grid, horizontal, list)

2. Skill Management

Click the 🛠️ icon on any agent card
Enable/disable skills per agent
Automatic blacklist/whitelist management
View all available skills

3. Subagent Dispatch

Click "调度管理" (Manage Dispatch)
Select a dispatcher agent
Configure allowed subagents
Set max concurrent tasks

4. Create New Agent

Click "新建 Agent" (Create Agent)
Configure agent ID, name, emoji, color theme
Select model provider and model
Set system prompt

5. Settings

Click "设置" (Settings)
Customize dashboard title and subtitle
Configure auto-refresh interval
Set token cost estimates
Show/hide cost information

Configuration

The dashboard reads configuration from:

~/.openclaw/openclaw.json - OpenClaw global config
~/.config/agent-dashboard/config.json - Dashboard settings (auto-created)

Environment Variables

OPENCLAW_HOME=/home/user/.openclaw  # Path to OpenClaw installation
DASHBOARD_PORT=5181                  # Dashboard server port

Architecture

Backend: Flask (Python 3)
Frontend: Vanilla JavaScript + CSS
Data Source: OpenClaw JSON configuration
Authentication: Inherits OpenClaw authentication

API Endpoints

The dashboard provides REST API endpoints:

GET  /api/agents              # List all agents
POST /api/agents              # Create new agent
GET  /api/agents/\x3Cname>       # Get agent details
DELETE /api/agents/\x3Cname>     # Delete agent
GET  /api/skills              # List all skills
GET  /api/agents/\x3Cname>/skills # Get agent skills
POST /api/agents/\x3Cname>/skills/\x3Cid>/enable
POST /api/agents/\x3Cname>/skills/\x3Cid>/disable
GET  /api/stats               # Get statistics

Requirements

Python 3.8+
OpenClaw installed and configured
Modern web browser

File Structure

agent-dashboard/
├── dashboard_server.py      # Flask backend
├── openclaw_config.py       # OpenClaw config manager
├── openclaw_skills.py       # Skills management
├── openclaw_finder.py       # OpenClaw path finder
├── openclaw_schema.py       # Config validation
├── static/
│   ├── css/style.css        # Styles
│   ├── js/app.js            # Frontend logic
│   └── js/i18n.js           # Internationalization
├── install.sh               # Installation script
└── SKILL.md                 # This file

Troubleshooting

Port Already in Use

# Find and kill process using port 5181
lsof -ti:5181 | xargs kill -9
# Or use a different port
DASHBOARD_PORT=8080 python3 dashboard_server.py

Cannot Find OpenClaw

# Set environment variable
export OPENCLAW_HOME=/path/to/.openclaw

Permission Denied

# Fix permissions
chmod +x install.sh
chmod 644 static/*

Updating

cd ~/.agents/skills/agent-dashboard-v2
git pull origin main
# Restart service if using systemd
systemctl --user restart agent-dashboard

Uninstallation

# Stop service
systemctl --user stop agent-dashboard
systemctl --user disable agent-dashboard

# Remove files
rm -rf ~/.agents/skills/agent-dashboard-v2
rm -f ~/.config/systemd/user/agent-dashboard.service

License

MIT

Author

@mikezhuyan

Support

For issues and feature requests, visit: https://github.com/mikezhuyan/agent-dashboard/issues

安全使用建议

Before installing/running this skill: - Review the code (especially dashboard_server.py, install.sh and start.sh) to confirm authentication and binding behavior. The shipped config defaults bind the server to 0.0.0.0 (exposed to network). - Expect the dashboard to read and write ~/.openclaw/openclaw.json and to create/modify files under ~/.openclaw/agents — back up that directory first and inspect openclaw.json for API keys you may not want exposed. - Do not run install.sh or enable the systemd service without reading it; install.sh will create a user systemd unit and may start the service. - If you want safer deployment: run the dashboard in a sandbox or container, change host binding to 127.0.0.1, restrict firewall rules, or require explicit authentication on the dashboard before exposing it to any network. - If you install, verify that REST endpoints require appropriate auth and that uploaded avatars or created agent files are stored where you expect. Given the missing declarations in the metadata (no config paths or env vars listed) and the dashboard's ability to modify local OpenClaw data and serve it over the network, treat this package as requiring manual review and limited deployment (suspicious until validated).

功能分析

Type: OpenClaw Skill Name: digital-staff-dashboard Version: 2.0.0 The bundle provides a management dashboard with high-privilege capabilities, including scripts that modify system firewall rules (open_port.sh) and a Flask backend that executes CLI commands via subprocess (dashboard_server.py). It performs extensive file operations in the user's home directory and contains numerous hardcoded absolute paths (e.g., '/home/mike/...' in openclaw_skills.py and data/config.json), which indicates poor sanitization and potential for unexpected behavior. While these features support the stated goal of agent management, the combination of broad filesystem access, CLI execution, and a network-accessible server (defaulting to 0.0.0.0) presents a significant attack surface without clear evidence of malicious intent.

能力标签

cryptorequires-oauth-token

能力评估

⚠ Purpose & Capability

The skill's code and SKILL.md align with a dashboard for OpenClaw (finding OpenClaw installs, listing/creating agents, managing skills). However the registry metadata declares no required config paths or credentials while the implementation clearly reads and writes ~/.openclaw and creates agent directories in ~/.openclaw/agents. That mismatch (no declared config paths but extensive file access) is an inconsistency the user should be aware of.

⚠ Instruction Scope

SKILL.md and the included scripts instruct the user to clone/run install.sh or start dashboard_server.py. The code searches the user's home for OpenClaw installations, reads ~/.openclaw/openclaw.json, and (per README and scripts) can create new agent directories and write model/metadata files into ~/.openclaw/agents. The dashboard exposes REST endpoints for listing/creating/deleting agents and uploading avatars. These operations touch and modify potentially sensitive local configuration (including model provider entries and API keys) and are broader than metadata claims.

ℹ Install Mechanism

There is no formal install spec in the registry metadata, but the package includes install.sh and start.sh. install.sh will pip install Flask (via pip --user), create a systemd user service file, enable it, and optionally start it. The install approach is typical but performs filesystem writes (systemd user unit, desktop shortcut) and launches a service — so the user should review the install script before running it.

⚠ Credentials

Registry metadata lists no required environment variables, but SKILL.md and scripts reference OPENCLAW_HOME and DASHBOARD_PORT. More importantly, the dashboard reads/writes ~/.openclaw/openclaw.json and agent model files; those files often contain model provider configurations and API keys. Requiring (and manipulating) these local config files is a legitimate need for a management dashboard, but the omission from declared requirements and the potential to copy/store provider API keys inside agent files is a sensitive capability that should be explicitly called out to users.

ℹ Persistence & Privilege

always is false. The installer (with interactive prompts) can create and enable a user-level systemd service that will autorun the dashboard. The dashboard defaults to binding to all addresses (config DEFAULT host is '0.0.0.0' and server.log shows 'Running on all addresses (0.0.0.0)'), which can expose the dashboard and any OpenClaw configuration it serves to the local network. This persistence and network exposure are significant and require user attention, but they are not set automatically by the registry metadata — they happen via the provided install script if the user consents.

版本历史

v2.0.0

**Major update: Agent Dashboard V2 is a modern, web-based dashboard for real-time OpenClaw agent management.** - Complete redesign with a modern web UI, dark theme, and real-time agent status/statistics. - Adds advanced features: skill management, token usage tracking, subagent dispatch, new agent creation, and multi-language support. - Introduces a Python3/Flask backend with REST API and a new static frontend (no external services required). - Replaces old scripts and templates with a full-featured, responsive web app; legacy Markdown setup instructions and sample data files removed. - Installation and configuration streamlined—supports systemd service, environment variables, and integrated troubleshooting. - Compatibility: requires OpenClaw, Python 3.8+, and a modern browser.

元数据

Slug digital-staff-dashboard

版本 2.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题

digital staff 是什么？

A modern web-based dashboard for managing OpenClaw agents with real-time monitoring, token usage tracking, skill management, and multi-language support. Prov... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 85 次。

如何安装 digital staff？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install digital-staff-dashboard」即可一键安装，无需额外配置。

digital staff 是免费的吗？

是的，digital staff 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。

digital staff 支持哪些平台？

digital staff 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 digital staff？

由 mikezhuyan（@mikezhuyan）开发并维护，当前版本 v2.0.0。

digital staff