customer-satisfaction
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection attacks by processing untrusted data from external customer feedback channels.
- Ingestion points: Untrusted data enters the agent context through the collection of post-resolution CSAT surveys, ticket sentiment analysis, and monitoring of public reviews from platforms like G2, Capterra, and TrustRadius (documented in SKILL.md).
- Boundary markers: The skill instructions do not specify the use of delimiters or explicit boundary markers to separate user/customer content from agent instructions during analysis.
- Capability inventory: The agent is tasked with sentiment analysis, theme extraction, and generating alerts based on the content of these external messages, which could be manipulated by malicious feedback content.
- Sanitization: There is no documentation or instruction for sanitizing, escaping, or validating the external feedback before it is processed by the AI agent.
Audit Metadata