Security audit

Markdown导出工具

Security checks across malware telemetry and agentic risk

Overview

The core Markdown conversion use is understandable, but the skill also introduces unclear callback, API, and credential behavior that users should review before installing.

Review this skill carefully before installing. Use it only for Markdown conversion tasks where you are comfortable with the agent reading input files, writing outputs, and running conversion commands. Do not provide API keys, credentials, callback URLs, or administrator privileges unless the publisher clearly explains the exact service, data sent, and reason those privileges are required.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (6)

Intent-Code Divergence

Medium

Confidence: 91% confidence
Finding: The skill claims file-path-oriented local processing, but also documents a callback_url for asynchronous notifications without clearly defining what data is sent, to whom, or under what trust boundary. In an agent context, this ambiguity can enable unintended outbound transmission of user content, file metadata, or execution status to external systems.

Context-Inappropriate Capability

Medium

Confidence: 89% confidence
Finding: The documentation introduces API integration, credential setup, and network error handling unrelated to a simple Markdown exporter's stated purpose. This expands the implied capability surface and can mislead an agent into requesting secrets or making unnecessary external connections, increasing exfiltration and misuse risk.

Intent-Code Divergence

Medium

Confidence: 93% confidence
Finding: The skill presents itself as a local conversion utility but later instructs users to configure API keys and establish API connections, creating a hidden trust-boundary shift. Such contradictions are dangerous because agents or users may expose credentials or route document content to external services under the false assumption the tool is purely local.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The manifest description lacks a concrete trigger list and leaves activation scope ambiguous, which is risky for agentic routing. Ambiguous scope can cause over-selection of this skill, exposing local files or enabling command execution in situations that do not actually require Markdown conversion.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The manifest description lacks a concrete trigger list and leaves activation scope ambiguous, which is risky for agentic routing. Ambiguous scope can cause over-selection of this skill, exposing local files or enabling command execution in situations that do not actually require Markdown conversion.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The documentation references callback URLs and API/external service behavior without warning that user content, filenames, metadata, or processing results may be transmitted off-system. In a document-processing skill, silent outbound transfer creates a meaningful confidentiality risk because users may assume their Markdown files remain local.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.