now-assist-qa

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill utilizes ServiceNow-specific MCP and REST tools to manage configuration records. All network requests are directed to relative ServiceNow API paths within the authenticated environment.
  • [SAFE]: Security best practices are integrated into the instructions, specifically regarding the configuration of safety guardrails, PII blocking, and knowledge grounding to prevent hallucinations and data exposure.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes data from external sources such as knowledge bases and user feedback records. 1. Ingestion points: ServiceNow kb_knowledge and sn_now_assist_feedback tables. 2. Boundary markers: The skill instructions include configuring system prompts that explicitly define assistant boundaries. 3. Capability inventory: Tools for querying, creating, and updating ServiceNow table records. 4. Sanitization: The procedure includes mandatory steps for setting up safety filters and confidence scores.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 02:24 PM
Security Audit — agent-trust-hub — now-assist-qa