PDF视觉阅读器

• Outcome: Pass
• Latest audit: May 29, 2026, 8:30 PM
• Version:
  1.0.0

Overview

This PDF reader does what it says, but users should only use it on documents they are comfortable converting to images and analyzing with their AI model.

Use this skill only for PDFs you are willing to have rendered into local image files and processed by your configured AI vision model. For confidential, legal, medical, financial, or proprietary PDFs, check your model provider's data handling terms, limit analysis to needed pages, and avoid memory archiving unless you want the derived content retained.

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 (5)

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The README explicitly states that PDF-derived page images are sent to an AI vision model, but it does not disclose whether those images leave the local environment, what provider receives them, or the privacy implications for sensitive PDF content. Because PDFs often contain confidential business, personal, or regulated data, this omission can lead users to unknowingly transmit sensitive information to third-party services.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The activation text is broad enough to trigger on many generic PDF-related requests, which can cause the skill to process documents in situations the user did not clearly intend. In this skill, broad activation matters because processing includes rendering full PDF pages and sending them to a vision model, potentially exposing sensitive document contents unnecessarily.

Vague Triggers

Medium

• Confidence: 86% confidence
• Finding: The usage scenarios are framed in vague terms like 'PDF needs analysis' and 'scan recognition,' without clear boundaries for when the skill should or should not run. That ambiguity increases the chance of over-collection and unintended use on sensitive PDFs when a safer or more limited workflow might suffice.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill describes converting PDF pages into images and submitting them to a vision AI model, but it gives no user-facing warning that full page contents may be transmitted to a third-party service. Because PDFs often contain sensitive personal, legal, financial, or confidential business data, silent external transmission creates a meaningful privacy and data-governance risk.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill proposes archiving derived content into a memory/ directory without clearly warning that document-derived information may be stored persistently. Persistent storage can extend the exposure window for sensitive information and create secondary risks such as unauthorized reuse, later disclosure, or retention-policy violations.

VirusTotal

65/65 vendors flagged this skill as clean.