openclaw-zero-token
Fail
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The installation process involves cloning a repository from a personal, unverified GitHub account (
github.com/linuxhsj/openclaw-zero-token) and installing its dependencies usingpnpm install. This introduces a significant supply chain risk. - [COMMAND_EXECUTION]: The skill provides an
exectool that grants AI models the capability to execute arbitrary shell commands on the user's machine. While the documentation mentions keyword triggers, this remains a high-risk capability. - [DATA_EXFILTRATION]: The core functionality involves capturing and storing highly sensitive authentication data, including session cookies and bearer tokens, from browser sessions for multiple providers (Claude, ChatGPT, DeepSeek, etc.). If the underlying downloaded code is compromised, these credentials could be exfiltrated.
- [COMMAND_EXECUTION]: The installation instructions for Node.js involve fetching a remote script from NodeSource and piping it directly to
sudo bash, which is a risky installation pattern. - [DATA_EXFILTRATION]: The tool calling feature includes
readandwritecapabilities. While documented as restricted to a workspace, these tools allow the model to access or modify local files.
Recommendations
- AI detected serious security threats
Audit Metadata