Trust Assessment
screen-vision received a trust score of 73/100, placing it in the Caution category. This skill has some security considerations that users should review before deployment.
SkillShield's automated analysis identified 2 findings: 0 critical, 2 high, 0 medium, and 0 low severity. Key findings include Skill captures entire screen, exposing all visible data, Skill can simulate mouse clicks at arbitrary screen coordinates.
The analysis covered 4 layers: Manifest Analysis, Static Code Analysis, Dependency Graph, LLM Behavioral Safety. All layers scored 70 or above, reflecting consistent security practices.
Last analyzed on February 13, 2026 (commit 13146e6a). SkillShield performs automated 4-layer security analysis on AI skills and MCP servers.
Layer Breakdown
Behavioral Risk Signals
Security Findings2
| Severity | Finding | Layer | Location | |
|---|---|---|---|---|
| HIGH | Skill captures entire screen, exposing all visible data The `scripts/vision_ocr.swift` script utilizes the `/usr/sbin/screencapture` utility to capture a screenshot of the entire user's screen. This grants the AI agent access to any information displayed on the screen at the time of execution, including potentially sensitive personal data, financial details, or private communications. While this is core to the skill's functionality, it represents a significant privacy risk if the agent is compromised or misused. Implement mechanisms to limit screen capture to specific application windows or regions, rather than the entire screen. Require explicit user confirmation for each capture, especially when sensitive applications are open. Ensure robust data handling and privacy policies for any captured content. | LLM | scripts/vision_ocr.swift:10 | |
| HIGH | Skill can simulate mouse clicks at arbitrary screen coordinates The `scripts/click.swift` script allows the AI agent to perform mouse clicks at any specified X, Y coordinates on the screen using CoreGraphics. If the coordinates are derived from untrusted user input without proper validation, or if the AI agent's decision-making process is compromised, this capability could be exploited to perform unintended or malicious actions on the user's system, such as clicking on 'confirm' buttons, deleting files, or interacting with sensitive applications. Implement strict validation and sanitization of input coordinates to prevent out-of-bounds or malicious clicks. Consider requiring explicit user confirmation for click actions, especially for sensitive operations. Limit the scope of clickable areas or provide a visual indicator of where a click will occur. | LLM | scripts/click.swift:6 |
Scan History
Embed Code
[](https://skillshield.io/report/446fdd9e6a2b59d8)
Powered by SkillShield