DELLIGHT CMO Content Marketing

• Outcome: Pass
• Latest audit: May 28, 2026 at 1:24 PM UTC
• Version:
  1.1.0

Overview

This is an instruction-only DELLIGHT.AI marketing skill that is coherent and does not show hidden execution, data theft, persistence, or destructive behavior.

Safe to install as a marketing playbook. Keep human review before publishing, replying, tagging people, reposting, using customer examples, or generating Arthur-style cloned voice, and treat third-party AI/media tools as external services that may receive the content you provide.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The activation text is broad enough to trigger on many generic marketing, copywriting, campaign, and planning requests, which can cause the skill to activate outside its intended scope. Over-broad activation increases the chance of prompt-context hijacking of unrelated user tasks and can silently steer outputs toward the skill's fixed business objectives, such as tying all activity to revenue generation.

Natural-Language Policy Violations

High

• Confidence: 90% confidence
• Finding: The rule 'All text/screens in readable English — no foreign characters' imposes an unconditional language restriction that can override user intent, exclude multilingual content, and bias outputs in ways unrelated to safety or functionality. In a marketing skill, this is more dangerous because it can systematically suppress localization, international audiences, and legitimate non-English brand or customer content.

VirusTotal

66/66 vendors flagged this skill as clean.