Document Parse Ocr

• Outcome: Pass
• Latest audit: Jun 3, 2026, 2:44 AM
• Version:
  1.0.0

Overview

This is a straightforward Scnet OCR skill that processes user-provided public document URLs, with privacy precautions needed for sensitive files.

Install only if you are comfortable sending the document URL and OCR output through Scnet. Use short-lived, access-scoped public URLs without embedded secrets, avoid confidential or regulated documents unless approved for third-party OCR processing, protect the SCNET_API_KEY, and do not change SCNET_API_BASE unless you trust the endpoint.

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 (4)

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The AI-trigger guidance is broad enough that an agent may invoke the skill whenever a user mentions parsing a document or supplies a URL, without confirming that the user wants third-party OCR processing. Because the skill sends document URLs and derived content to an external service, ambiguous triggering increases the risk of unintended data disclosure.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill explains how to obtain and use the OCR service but does not prominently warn that submitted document URLs and document contents are transmitted to a third-party provider for processing. In this context, the missing privacy notice is significant because users may unknowingly expose confidential documents, contracts, or regulated data to an external processor.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill requires files to be hosted at publicly accessible URLs but does not clearly warn that making documents public can itself create a data exposure event, even before OCR begins. This is especially dangerous for internal, personal, legal, or financial documents, where temporary public hosting may be indexed, logged, or accessed by unintended parties.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The documentation instructs users to submit publicly accessible file URLs to a third-party OCR service and use bearer-authenticated API calls, but it does not warn that documents may contain sensitive data or that making files publicly reachable can expose them to unintended access. In this skill context, the service processes full document contents including tables, headers, footnotes, stamps, and markdown text, which increases the privacy and confidentiality risk for enterprise documents.

VirusTotal

65/65 vendors flagged this skill as clean.