tzai-empty-state
Warn
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill constructs and executes a shell command in
SKILL.mdthat directly interpolates user-provided text. If this input is not properly escaped by the agent, it could allow an attacker to execute arbitrary commands via shell metacharacters. Evidence:bash "$ENGINE" empty-state --prompt "<user subject>". - [COMMAND_EXECUTION]: The skill in
SKILL.mddynamically resolves the path to an executable by checking several hidden directories in the user's home folder ($HOME/.agents, $HOME/.claude, etc.). Evidence: The 'Resolve engine' script block iterates through multiple home directory paths to define the$ENGINEvariable.
Audit Metadata