ab-test-setup
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides structured instructions and best practices for experimentation and A/B testing methodology without any malicious patterns or risky operations.\n- [NO_CODE]: The skill consists entirely of Markdown files and lacks executable scripts or code, precluding risks like remote code execution or privilege escalation.\n- [SAFE]: All referenced external tools, including Evan Miller's calculator, Optimizely, and VWO, are well-known and reputable services in the experimentation industry.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes a surface for ingesting untrusted data from a local context file, although it possesses no exploitable capabilities.\n
- Ingestion points: The skill instructions in
SKILL.mddirect the agent to read.claude/product-marketing-context.mdif it exists.\n - Boundary markers: No specific boundary markers or delimiters are defined for the ingested context.\n
- Capability inventory: The skill does not contain any tool definitions, subprocess calls, file-write operations, or network exfiltration patterns.\n
- Sanitization: No sanitization or filtering is applied to the ingested content.
Audit Metadata