kb-summarization
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions.
- Ingestion points: The skill retrieves potentially untrusted content from the
textandwikifields of thekb_knowledgetable, as well as user-submittedcommentsfrom thekb_feedbacktable inSKILL.md(Step 1, Step 3). - Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potential commands embedded within the article text or user feedback during the summarization steps.
- Capability inventory: The skill is granted access to ServiceNow modification tools (
SN-Update-Record) and nativeBashexecution, which increases the impact if an injection were to be successful. - Sanitization: No sanitization or validation logic is defined to filter out prompt injection patterns from the article content or feedback data before it is processed by the model.
Audit Metadata