article-generation

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's behavior is entirely consistent with its described purpose of generating knowledge articles from ServiceNow record data.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from ServiceNow records (e.g., incident description, close notes) which could theoretically contain prompt injections. However, the risk is mitigated by the workflow requiring articles to be created in a 'draft' state for human review.
  • Ingestion points: ServiceNow incident, problem, and change_request tables (SKILL.md, Step 1, 7, 8).
  • Boundary markers: The instructions direct the agent to map data into a specific HTML structure (SKILL.md, Step 4), providing implicit structural boundaries.
  • Capability inventory: Record creation/update permissions on ServiceNow tables and work note logging (SKILL.md, tools section).
  • Sanitization: No automated sanitization is specified, but the procedure enforces a 'draft' workflow for manual verification.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 01:34 PM
Security Audit — agent-trust-hub — article-generation