churn-risk-detection
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists exclusively of markdown instructions and metadata. It does not include any executable scripts, shell commands, or external tool definitions. The methodology is designed for purely instructional use within an agent's reasoning process.- [SAFE]: Comprehensive analysis of the methodology and metadata revealed no malicious patterns, such as obfuscation, credential theft, or unauthorized network activity. The skill adheres to its stated purpose without hidden behaviors.- [PROMPT_INJECTION]: An evaluation of potential indirect prompt injection was performed. Although the skill identifies 'Transaction Data' and 'Customer IDs' as ingestion points (SKILL.md) and lacks explicit delimiters or sanitization, it possesses no execution capabilities or system access tools, neutralizing any risk associated with processing untrusted data.
Audit Metadata