audit-automation-gaps
Warn
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill employs dynamic context injection in
SKILL.mdusing the!bash ...syntax to executescripts/inventory.shautomatically upon skill initialization. While the script currently performs local configuration inventory, this mechanism allows for silent code execution at load time. - [COMMAND_EXECUTION]: In Phase 2.1 of
SKILL.md, the instructions direct the agent to executetime <tool-command>. Because the command name is derived from a list of candidates generated by scanning the repository, it creates a surface for command injection. Malicious repository content (e.g., file names or logs) could be crafted to include shell metacharacters that would be executed during the performance measurement phase.
Audit Metadata