ai-lens
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted record data from ServiceNow without sanitization or boundary markers, creating a surface for indirect prompt injection. \n
- Ingestion points: User-supplied record fields such as short_description and description are ingested from ServiceNow tables (e.g., incident, sn_customerservice_case, sn_hr_core_case) using tools like SN-Get-Record and SN-Query-Table. \n
- Boundary markers: The procedures do not include delimiters or specific instructions for the agent to ignore or isolate commands embedded within the retrieved data. \n
- Capability inventory: The skill provides the agent with capabilities to query system data (SN-Query-Table), search knowledge bases (SN-NL-Search), and modify system records (SN-Update-Record, SN-Create-Record). \n
- Sanitization: No escaping, validation, or filtering is applied to the record content before it is interpolated into prompts or search queries.
Audit Metadata