returns-reverse-logistics
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted customer input, such as return reasons and defect statements, which are interpolated into communication templates. While this creates an indirect injection surface, the risk is mitigated by instructions to confirm all actions from system evidence and the lack of autonomous execution capabilities.\n
- Ingestion points:
references/communication-templates.md(variables including{{return_reason}},{{defect_description}}, and{{customer_defect_statement}}).\n - Boundary markers: The templates do not utilize explicit delimiters or guardrail instructions for the untrusted variables.\n
- Capability inventory: The skill is restricted to drafting content; it explicitly prohibits unauthorized financial or inventory actions.\n
- Sanitization: Not provided for string inputs, although comprehensive for physical media sanitization.\n- [SAFE]: The skill implements significant data protection measures, including instructions to use masked lookups for payment data and strict prohibitions against inspecting customer files on returned devices.
Audit Metadata