ab-testing
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows security best practices for A/B testing implementation and does not contain malicious code or instructions.
- [EXTERNAL_DOWNLOADS]: The skill references other utilities from the same author, such as
conversion-analyticsandux-writer, which are installed using a platform-specific command (npx skills add). These are legitimate vendor resources within the author's ecosystem. - [DATA_EXPOSURE]: The instructions involve reading project configuration files (e.g.,
package.json,next.config.js) and source code to adapt the A/B testing implementation to the local stack. This is a functional requirement for generating relevant code variants. - [COMMAND_EXECUTION]: The skill instructs the agent to execute standard development commands and platform-specific skill management tools. No unauthorized or high-privilege command execution was detected.
- [INDIRECT_PROMPT_INJECTION]: The skill's workflow involves processing existing project code to design variants. While this is an ingestion point for untrusted data, the risk is mitigated by a 'pre-flight review' step where the agent audits variants for consistency and errors before deployment.
Audit Metadata