channel-performance-review
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill instructions define a clear analytical persona and specific domain rules. There are no attempts to bypass safety filters, extract system prompts, or override core agent instructions.
- [DATA_EXFILTRATION]: The skill uses platform-specific tools to access store data but contains explicit prohibitions against exposing customer PII (names, emails, addresses, phone numbers). No unauthorized network calls or credential harvesting patterns were observed.
- [REMOTE_CODE_EXECUTION]: The skill does not use any external dependencies, remote scripts, or dynamic code execution methods such as eval() or exec().
- [COMMAND_EXECUTION]: There is no evidence of shell command execution, privilege escalation (sudo), or unauthorized file system modifications.
- [SAFE]: The skill follows least-privilege principles by using aggregated data for its primary analysis and requires explicit user intent for order-level detail.
Audit Metadata