Security audit

图解

Security checks across malware telemetry and agentic risk

Overview

This Mermaid diagram skill is mostly documentation-only, but it asks for broad command execution and describes generic API and file operations beyond its diagram purpose.

Review this skill before installing. It may be acceptable only if you intend to let a diagram helper read/write files and run narrowly controlled rendering commands. Avoid granting broad shell access or API credentials unless the publisher documents exact commands, endpoints, and data handling boundaries.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill declares the `exec` tool even though the documented purpose is Mermaid diagram generation, which normally only requires text transformation and possibly file write/read access. Unnecessary command execution expands the attack surface significantly because prompt-controlled inputs could be routed into shell commands by the agent runtime or future skill logic, enabling arbitrary local actions unrelated to diagram rendering.

Context-Inappropriate Capability

Medium

Confidence: 84% confidence
Finding: Advertising generic external API integration as a core feature is not justified by the stated scope of a Mermaid diagram skill and signals potential over-privilege or scope creep. Even if no concrete API call is shown here, encouraging unrelated outbound connectivity can enable data exfiltration or unauthorized transmission of user content if implemented by the agent.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: Listing command execution as a core feature normalizes behavior that is outside the justified scope of diagram generation and can mislead users or orchestration systems into granting excessive privileges. In a skill that processes user-provided content, this creates a pathway for prompt-to-command abuse if future implementations invoke shell commands based on that content.

Description-Behavior Mismatch

Medium

Confidence: 92% confidence
Finding: The documentation broadens the skill from Mermaid generation into general file handling, API usage, and command execution, which is inconsistent with least privilege. This mismatch makes the skill more dangerous in context because a seemingly simple visualization skill may be approved with capabilities that permit filesystem access, network egress, and local command execution.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.