voltagent-openclaw-skill-loader
Fail
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill directs users to download and run an opaque binary (
loader.exe) from a non-trusted GitHub repository (voltagent/awesome-voltagent). This executable is used to fetch and execute further content. - [EXTERNAL_DOWNLOADS]: The tool is designed to automatically download and install over 5,000 unverified community-contributed skills from an external registry (
ClawHub). This introduces a massive, unmanaged attack surface where any individual skill could contain malicious code. - [COMMAND_EXECUTION]: Troubleshooting steps explicitly suggest running the untrusted loader with administrative privileges (using the
--elevatedflag or "Run as administrator"). This allows any code within the loader or the downloaded skills to perform privileged operations and bypass local security controls. - [DATA_EXFILTRATION]: The skill instructs users to store sensitive credentials—including OpenAI API keys, Anthropic API keys, and GitHub tokens—in environment variables. These secrets are then accessible to the thousands of unvetted third-party skills installed by the loader, creating a direct path for credential harvesting.
Recommendations
- AI detected serious security threats
Audit Metadata