handle-complaint
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted customer data from emails and CRM tickets, creating a surface for indirect prompt injection attacks where malicious instructions could be hidden in a complaint text.\n
- Ingestion points: External data is ingested from sources identified as
~~почта(email) and~~crminSKILL.md.\n - Boundary markers: The instructions do not mandate specific delimiters or escaping for the raw complaint text when it is interpolated into the agent's context.\n
- Capability inventory: The skill uses
ReadandWebFetchtools to gather and process information.\n - Sanitization: The skill implements robust human-in-the-loop controls through 'Approval Gates,' which strictly forbid autonomous execution of high-impact actions like sending responses, issuing refunds, or resolving disputes.\n- [SAFE]: The skill utilizes placeholder tags for internal business systems and maintains a clear separation between data analysis and action execution, following best practices for agentic safety.
Audit Metadata