hrsd-sentiment-analysis

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes free-text data from potentially adversarial sources.
  • Ingestion points: The skill reads from sys_journal_field (comments and work notes), interaction (chat history), and asmt_assessment_instance_question (survey responses).
  • Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the processed text to prevent the LLM from obeying instructions inside the data.
  • Capability inventory: The agent has the capability to modify records and add work notes via SN-Add-Work-Notes and SN-Update-Record tools based on its analysis.
  • Sanitization: No mention of sanitizing or validating external content before sentiment analysis is performed.
  • [DATA_EXFILTRATION]: The skill accesses and evaluates highly sensitive employee PII and HR case history.
  • Evidence: The skill queries sn_hr_core_profile for tenure, review dates, and job titles, and sn_hr_core_case for detailed workplace complaints and grievances. While authorized for the stated HR role, the aggregation of this data for automated risk assessment constitutes a high-sensitivity data handling scenario.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 02:24 PM
Security Audit — agent-trust-hub — hrsd-sentiment-analysis