design-doc-generator

• Outcome: Review
• Latest audit: May 29, 2026 at 9:29 PM UTC
• Version:
  1.0.0

Overview

This skill has a legitimate documentation purpose, but it asks for raw application login credentials and saves authenticated screenshots and internal design material without clear safeguards.

Install only if you are comfortable with the agent reading the relevant frontend and backend code and saving local screenshots, notes, and a Word document. Use a temporary or least-privileged test account, avoid sharing production passwords in chat, and review/redact screenshots and generated documents before distributing them.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (2)

Vague Triggers

Medium

• Confidence: 85% confidence
• Finding: The activation criteria are broad enough to trigger on ordinary documentation-related requests, which can cause the agent to begin extensive code review, site navigation, and file generation when the user may have intended a lightweight answer. In this context, overbroad triggering is risky because the skill performs high-impact actions such as reading project code and accessing live pages.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill explicitly asks for tenant, username, and password to log into the frontend application, but provides no safeguards, minimization rules, or warning about sensitive credential handling. This creates a direct path for collecting secrets through the agent workflow and increases the chance of credential exposure in prompts, logs, screenshots, notes, or generated documents.

VirusTotal

66/66 vendors flagged this skill as clean.