skills/insight68/skills/ab-test-setup/Gen Agent Trust Hub

ab-test-setup

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFENO_CODEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns, encoded commands, or deceptive metadata were detected. The skill's behavior aligns with its stated purpose of experiment design.\n- [NO_CODE]: The skill consists exclusively of Markdown files. It does not contain any scripts, binaries, or instructions for system modification.\n- [EXTERNAL_DOWNLOADS]: Documentation includes links to established and well-known A/B testing calculators (Optimizely, VWO, Evan Miller). These are neutral references for educational purposes and do not trigger remote code execution.\n- [PROMPT_INJECTION]: No attempts to bypass safety filters or override system prompts were found. The skill includes an indirect prompt injection surface which is considered low risk given the lack of executable capabilities:\n
  • Ingestion points: .claude/product-marketing-context.md (referenced in SKILL.md)\n
  • Boundary markers: Absent\n
  • Capability inventory: No scripts or tool-use capabilities are defined within this skill.\n
  • Sanitization: Absent
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 08:21 PM