command-creator
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [Command Execution] (SAFE): The skill demonstrates how to use the Bash tool (e.g., !
git status) within command templates. This is a documented feature for gathering repository context and the skill emphasizes using the 'allowed-tools' property to restrict permissions as a security measure.\n- [Indirect Prompt Injection] (LOW): The skill templates interpolate user-provided arguments ($1, $ARGUMENTS) directly into prompts. Evidence Chain: 1. Ingestion point: User-provided slash command arguments; 2. Boundary markers: Absent in the documentation examples; 3. Capability inventory: Bash, Read, Write, and Edit tools; 4. Sanitization: None specified in the templates. While this creates a surface for indirect prompt injection, it is the intended functional mechanism for passing parameters to custom commands.
Audit Metadata