document-management

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill contains a template processing engine in SKILL.md (functions processTemplateVariables and getFieldValue) that exhibits a surface for indirect prompt injection.
  • Ingestion points: Data is ingested from ServiceNow record fields (e.g., incident.short_description) and attachment records.
  • Boundary markers: The interpolation logic uses ${field} syntax but lacks boundary markers or instructions to the agent to ignore embedded commands within the data.
  • Capability inventory: The skill has access to high-privilege tools such as snow_execute_script_with_output and snow_query_table via the platform configuration.
  • Sanitization: The implementation lacks sanitization or HTML escaping for values retrieved from getDisplayValue() before they are interpolated into the document content.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 03:35 PM
Security Audit — agent-trust-hub — document-management