openclaw-master-skills
Warn
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation provides instructions to download a large collection of skills from an external GitHub repository (LeoYeAI/openclaw-master-skills) and via a custom installation tool (clawhub).- [COMMAND_EXECUTION]: The skill includes instructions for various shell commands, including git clone for remote code retrieval, cp for file system operations, and chmod for modifying file permissions. It also describes running package managers like pip and npm.- [REMOTE_CODE_EXECUTION]: The skill facilitates the installation and execution of a massive library of unverified third-party scripts. This process involves installing dependencies and executing code from over 1200 different modules across various languages like Python and JavaScript.- [PROMPT_INJECTION]: The indexing scripts in SKILL.md ingest data from external README files and metadata without sanitization or boundary markers, creating an indirect prompt injection surface where untrusted content within the skill library could influence agent behavior during discovery.
Audit Metadata