hrsd-chat-reply-recommendation

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates untrusted external data into its response generation process.
  • Ingestion points: Untrusted data enters the context via chat messages from sys_cs_message and article content from kb_knowledge.
  • Boundary markers: There are no defined delimiters or instructions to ignore embedded commands within the prompt logic described in Step 7.
  • Capability inventory: The skill can perform write operations to ServiceNow via SN-Add-Work-Notes and has access to the Bash tool.
  • Sanitization: No data validation, escaping, or sanitization steps are documented for the retrieved text before it is used to generate recommendations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 02:23 PM
Security Audit — agent-trust-hub — hrsd-chat-reply-recommendation