generate-talking-points
Warn
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
SN-Execute-Background-Scripttool in Step 4 to run arbitrary server-side JavaScript on the connected ServiceNow instance. This capability allows the execution of logic that bypasses standard API constraints. - [DATA_EXFILTRATION]: The skill is designed to access and aggregate highly sensitive Personal Identifiable Information (PII) and performance data, including:
- Talent profiles (
sn_hr_talent_profile): Skills, competencies, and risk of loss. - Review records (
sn_hr_talent_review): Historical performance ratings and manager comments. - Feedback records (
sn_hr_talent_feedback): Peer-provided feedback and ratings. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from the
sn_hr_talent_feedbacktable which is populated by other employees. - Ingestion points: Data is pulled from the
contentfield of thesn_hr_talent_feedbacktable in Step 3 and Step 5. - Boundary markers: The skill lacks explicit delimiters or instructions for the agent to ignore potential malicious commands embedded within the feedback text.
- Capability inventory: The skill has access to
SN-Execute-Background-Script,SN-Query-Table, andBashtools, providing a significant impact surface if an injection is successful. - Sanitization: There is no evidence of semantic sanitization or filtering to prevent the agent from obeying instructions hidden within the peer feedback strings.
Audit Metadata