customer-churn-signals
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies and processes external datasets, including churned-account lists, CRM activity logs, and product event streams. This creates a surface for indirect prompt injection if the ingested data contains instructions designed to influence the agent's behavior.
- Ingestion points: Data entered by the user during the 'Interview' phase and data processed during the backtesting 'Workflow' (SKILL.md).
- Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the analyzed data.
- Capability inventory: The skill is restricted to analytical tasks, generating signal definitions, and outputting structured reports or queries. It does not have capabilities for direct network operations, file-system modification, or shell execution.
- Sanitization: No sanitization or content validation for the input data is specified.
- [SAFE]: The skill does not contain any evidence of credential exfiltration, obfuscation, unauthorized remote code execution, or persistence mechanisms. All referenced external resources, such as the
mbfinotti/revops-skills@customer-health-scoreskill, are associated with the same author and are used for legitimate cross-skill functionality.
Audit Metadata