audit-perf
Warn
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The SKILL.md file contains a dynamic context injection pattern (
!cat ...) that executes a shell command during the skill loading phase, allowing for automated execution without user oversight. - [DATA_EXFILTRATION]: The command specifically targets the user's home directory to read
~/.claude/skills/audit-workflow.md. This pattern exposes local file contents to the LLM's context, which could be used to harvest configuration data or other skill instructions stored on the host machine.
Audit Metadata