resolve-customer-support-case
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted customer data, which creates a vulnerability surface where a user could potentially manipulate the agent's behavior through embedded instructions.\n
- Ingestion points: The skill records the customer's description of events and expected outcomes in
references/case-decisions-remedies-and-playbooks.md.\n - Boundary markers: There are no explicit instructions to use delimiters or specific boundary markers to separate customer input from the system's instructions.\n
- Capability inventory: The skill authorizes high-impact actions including issuing refunds, waiving fees, and granting policy exceptions.\n
- Sanitization: The provided files do not mention any input validation, sanitization, or filtering of customer-provided text.
Audit Metadata