ab-test-setup
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured guidance for A/B testing and includes a benign Python script (
sample_size_calculator.py) that uses only the Python standard library to perform statistical calculations. - [DATA_EXFILTRATION]: No sensitive data access or exfiltration patterns detected. The external URLs provided in the documentation refer to well-known, industry-standard A/B testing tools and calculators (e.g., Evan Miller, Optimizely, VWO).
- [INDIRECT_PROMPT_INJECTION]: The skill specifies reading a local context file (
.claude/product-marketing-context.md). While this represents a data ingestion surface, the skill lacks exploitable capabilities like network operations or privileged file system access, rendering the risk negligible. - [COMMAND_EXECUTION]: The included Python script is intended for local execution by the agent to support the primary task. Analysis confirms it carries no dangerous side effects and performs only mathematical operations and basic CLI input/output.
Audit Metadata