customer-health

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] The skill processes data from untrusted external sources, including emails, chat transcripts, and external documents. While it implements a 'proposal-only' mode for content-originated actions to prevent unauthorized execution during unattended runs, the underlying risk of instructions embedded in those documents influencing the agent's reasoning remains a factor common to information-retrieval skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:04 PM
Security Audit — agent-trust-hub — customer-health