Security audit

节点

Security checks across malware telemetry and agentic risk

Overview

This skill is a broad connection-diagnostics template that asks for command execution and file-writing authority without clearly limiting what it may run or change.

Review this skill before installing. It may be reasonable only in a tightly supervised diagnostic session where you approve every command and file change. Do not give it access to sensitive projects, credentials, or production configuration unless the publisher narrows the command list, write targets, API usage, and repair behavior.

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 (5)

Intent-Code Divergence

Medium

Confidence: 86% confidence
Finding: The documentation claims the skill configuration is fixed and not user-customizable, yet earlier sections expose configurable behavior such as mode, max_retries, and skip_steps. This mismatch can mislead operators and downstream agents about the skill's true control surface, increasing the chance that execution-affecting inputs are accepted without proper review.

Intent-Code Divergence

Medium

Confidence: 93% confidence
Finding: The security section claims only whitelisted commands are executed, but the skill elsewhere advertises generic command execution without naming any whitelist or constraints. This creates a dangerous trust gap: users may assume command execution is restricted when the documented capability appears broad enough to permit arbitrary system commands.

Description-Behavior Mismatch

High

Confidence: 96% confidence
Finding: The manifest markets a narrow node connection diagnostic skill, but the description expands its use to broad API integration, interface docking, webhook configuration, and system connectivity tasks. This scope inflation is dangerous because it normalizes use of a narrowly trusted skill for much more powerful and less reviewable actions, including external integrations and system-level operations.

Context-Inappropriate Capability

Medium

Confidence: 88% confidence
Finding: A generic file write capability is not justified by a node-connection diagnostic purpose and expands the skill from observation into modification. If exposed to untrusted prompts or automation, this could overwrite configuration, persist unsafe artifacts, or alter files outside the expected diagnostic workflow.

Context-Inappropriate Capability

High

Confidence: 97% confidence
Finding: Generic system command execution is highly sensitive and is not justified by the skill's stated purpose of diagnosing node connection and pairing failures. In the context of an agent skill, such a broadly described exec capability can be repurposed for arbitrary command execution, host inspection, network access, or persistence, making the context substantially more dangerous.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.