setting-up-ab-tests
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists of benign educational content regarding marketing experimentation, statistical significance, and hypothesis testing.
- [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface by ingesting untrusted user input describing test contexts and conversion data. Ingestion points include the 'Initial Assessment' and 'Task-Specific Questions' sections. However, the risk is negligible as the skill lacks exploitable capabilities like shell execution, file writing, or network operations, and it defines clear structural boundaries for the hypothesis framework.
- [DATA_EXPOSURE]: The skill instructs the agent to read
.claude/product-marketing-context.mdif available. This is a local project context file intended to improve the relevance of the agent's advice and does not involve access to sensitive system credentials or private keys. - [EXTERNAL_DOWNLOADS]: The skill references official calculators from well-known services, including Evan Miller and Optimizely. These links are provided for user reference and represent standard tool integrations in this domain.
Audit Metadata