OCR Local V2

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:26 PM UTC
• Version:
  1.0.0

Overview

This OCR skill appears to do what it claims, with normal dependency and first-run language-data download caveats.

Install only if you are comfortable with npm fetching tesseract.js and with the first OCR run downloading and caching language data. For locked-down or offline environments, pre-vendor or pre-cache the dependency and language files, and use the skill only on images whose extracted text you are willing to show in your local agent session.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Description-Behavior Mismatch

Medium

• Confidence: 82% confidence
• Finding: The README markets the skill as '100% local' and 'No API key required', but also notes that the first run downloads Tesseract language data. This is a real security-relevant documentation issue because users may rely on the claim to use the skill in offline, restricted, or privacy-sensitive environments, while the implementation performs unexpected network access.

Intent-Code Divergence

Low

• Confidence: 76% confidence
• Finding: Saying the skill 'runs locally' is somewhat misleading when first-run behavior includes downloading language data. While the OCR processing itself may be local, the undocumented or under-emphasized network dependency can violate user expectations and security policies in locked-down environments.

Description-Behavior Mismatch

Medium

• Confidence: 91% confidence
• Finding: The skill advertises itself as "100% local" and "no API key required," but the notes explicitly state that the first run downloads language data. That creates a documentation integrity issue: users may assume no network access occurs, which can violate offline-only, privacy-sensitive, or supply-chain-restricted environments even if the OCR processing itself is local.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

"license": "MIT",
  "acceptLicenseTerms": true,
  "dependencies": {
    "tesseract.js": "^7.0.0"
  },
  "repository": {
    "type": "git",

• Confidence: 89% confidence
• Finding: "tesseract.js": "^7.0.0"

VirusTotal

66/66 vendors flagged this skill as clean.