teamclawtestv101

• Analyzed: May 11, 2026, 5:09 AM
• Findings: 1
• Version: 1.0.0
• Hash: 7d01c2061de97846256ef9d04aa738d5b2e0c6a8efdf31eac50a5edd648ceeb6
• Source: palm
• External report: View on VirusTotal

Findings (1)

The OpenClaw AgentSkills skill bundle "teamclawtestv101" implements a complex multi-agent system called Mini TimeBot. The system includes a core AI agent, an OASIS multi-expert orchestration engine, various MCP (Model Context Protocol) tool services, and integrations with external chat platforms (Telegram, QQ). **Analysis of Key Indicators:** 1. **Data Exfiltration / Unauthorized Remote Control:** * **`oasis/server.py`**: The OASIS service's `_run_discussion` function makes an `httpx.AsyncClient().post` call to a `callback_url` provided in the `CreateTopicRequest`. This `callback_url` can be an arbitrary URL. This allows the OASIS service to make outbound HTTP POST requests to any external endpoint, potentially exfiltrating data (like discussion conclusions) or triggering actions on other services. The `X-Internal-Token` is included in this callback, which could be leaked if the `callback_url` points to an attacker-controlled server. This is a **critical vulnerability** for Server-Side Request Forgery (SSRF) and data exfiltration. * **`chatbot/QQbot.py`**: Monkey-patches `aiohttp.ClientSession.__init__` to force all internal requests through a hardcoded SOCKS5 proxy at `127.0.0.1:1080`. While currently local, this is an unusual and risky modification of fundamental network behavior. If the proxy address could be controlled, it could be used for