TeamClaw

What This Skill Does

TeamClaw is a robust, high-performance orchestration subsystem for the OpenClaw ecosystem, designed to manage complex multi-agent workflows through its proprietary OASIS visual canvas. Unlike standard agents, TeamClaw acts as a parallel execution environment, allowing for computer-use tasks, scheduled automation via APScheduler, and real-time interactive monitoring. It provides a comprehensive Web UI that can be securely exposed via Cloudflare Tunnel, enabling remote control from mobile devices or workstations. The skill integrates seamlessly with social platforms like Telegram and QQ, turning your AI agents into active participants in your daily communication channels.

Installation

Installation follows the OpenClaw hub protocol. Run: clawhub install openclaw/skills/skills/avalon-467/teamclaw. Once installed, navigate to the project directory and execute the three-step lifecycle: first, run bash selfskill/scripts/run.sh setup to resolve dependencies. Second, run bash selfskill/scripts/run.sh configure --init followed by --batch arguments to define your LLM credentials. Finally, perform the critical step of user provisioning with bash selfskill/scripts/run.sh add-user <username> <password>. The system will not function without a designated user account, as all UI and API access is secured via authentication credentials.

Use Cases

TeamClaw is ideal for users requiring long-running, autonomous background tasks that do not disrupt their main agent workspace. Primary use cases include automated browser-based data extraction, periodic monitoring of web services, and creating 'agent teams' where multiple specialized models collaborate to solve a single complex prompt. It is also perfect for mobile power users who need to trigger workflows or check agent statuses while away from their desktop, thanks to the integrated Web UI and remote-access support.

Example Prompts

1. "TeamClaw, initiate a new OASIS workflow that scrapes the latest tech headlines from Reddit and synthesizes them into a summary for my Telegram channel."
2. "Set up a scheduled task in the system to check my company dashboard every hour and push a Bark notification to my phone if any errors appear."
3. "Orchestrate a multi-agent discussion between my CodingAgent and ReviewerAgent to debug the script in my workspace and present the final fix on the Web UI."

Tips & Limitations

Always remember that TeamClaw runs as a separate process; if the main OpenClaw agent crashes, the TeamClaw subsystem may persist, consuming system resources. Use status frequently to monitor process health. Security is paramount; ensure you do not use weak passwords during the add-user phase, as the Web UI is susceptible to unauthorized access if exposed via Cloudflare without proper authentication. Avoid running high-frequency automation tasks without configuring sufficient rate limits in your LLM provider settings to prevent unnecessary costs.