Diagram Generator

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:40 PM UTC
• Version:
  1.0.0

Overview

This is a disclosed AI diagram tool, with the main risks being external processing of selected diagrams and in-place edits that can overwrite files.

Install only if you are comfortable using an EvoLink API key and sending selected diagram prompts or files to EvoLink for AI processing. Avoid confidential architecture, network, database, or business diagrams unless EvoLink's data handling is acceptable to you, and keep backups or version control before using the edit command because it overwrites the chosen file.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (8)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 88% confidence
• Finding: The skill advertises and relies on capabilities including environment-variable access, shell execution, and reading/writing local files, but does not declare any permissions boundary for them. This creates a transparency and policy-enforcement gap: users or hosting platforms may treat the skill as lower risk than it is, while the documented commands can transmit local diagram content and API credentials may be used during shell-based operations.

Description-Behavior Mismatch

Medium

• Confidence: 92% confidence
• Finding: The script is explicitly designed to send prompts and diagram contents to EvoLink's remote API, and preview functionality can also open external services. This is a real security/privacy concern because users may provide proprietary diagrams or sensitive architecture details, and the network transmission behavior is broader than simple local diagram manipulation.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The PlantUML preview path encodes the full diagram and sends it to plantuml.com for rendering, disclosing diagram contents to a third party. This is dangerous when diagrams contain internal architecture, credentials, hostnames, or other sensitive metadata that users may not realize are leaving the local machine.

Context-Inappropriate Capability

Medium

• Confidence: 88% confidence
• Finding: The preview flow launches browser windows and, for some formats, directs users to external web services. This can cause unintended data exposure because diagram contents may be sent to third-party sites or opened in a browser context without an explicit consent step; in a security-sensitive environment, diagrams often contain internal architecture or secrets.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: AI commands send user descriptions and full file contents to a remote API without an in-flow disclosure warning at the point of transmission. That creates a meaningful confidentiality risk because users may unknowingly upload sensitive diagrams, schemas, or infrastructure details to an external provider.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The edit command replaces the original file directly with AI-generated output, which can cause irreversible loss or corruption of user data if the model returns malformed or incomplete content. Because the source file may be important documentation or system design material, silent overwrite increases operational risk.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The script transmits prompts and full diagram/file contents to an external API for generate, edit, convert, and explain operations without an in-band warning at the point of use. This is dangerous because users may process sensitive architecture, network, database, or UML files and unintentionally exfiltrate confidential information to a third party.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The edit command overwrites the original file directly with AI-generated output, with no backup, diff, or confirmation. This can destroy user data, corrupt diagram files, or silently replace trusted content with malformed or unsafe output, which is especially risky when editing important design artifacts.

VirusTotal

67/67 vendors flagged this skill as clean.