Security audit

DeepContent

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent DeepContent content-generation helper, with the main caution that it may search the public web for logos, headshots, and templates.

Install only if you trust the DeepContent MCP server and are comfortable with public web lookups for campaign assets. Prefer providing your own authorized image and logo URLs, and review any found assets for identity accuracy, source trust, and reuse rights before publishing.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

Medium

Confidence: 92% confidence
Finding: The skill says to use only MCP tool calls, but later instructs the agent to search the web for missing assets, creating an inconsistent execution boundary. That ambiguity can cause the agent to perform unscoped external browsing and fetch third-party URLs, which expands data exposure and supply-chain risk beyond the declared DeepContent server usage.

Description-Behavior Mismatch

Medium

Confidence: 86% confidence
Finding: The skill expands from DeepContent recipe/content generation into general web asset discovery, which is outside the narrowly described tool role. This broader behavior increases the chance of collecting untrusted external content or accessing unrelated resources under the guise of a content-generation request.

Vague Triggers

Medium

Confidence: 80% confidence
Finding: The trigger phrases are broad and overlap with common user intents such as 'generate content' or 'create a post,' which can cause unintended activation. Over-broad activation matters here because the skill includes external asset lookup and content-generation actions, so accidental invocation could lead to unnecessary data handling or tool use.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The skill instructs the agent to search online for headshots and logos when the user does not provide URLs, but it gives no privacy disclosure or consent step. This can expose personal names, affiliations, or event context to external services and may retrieve inaccurate or unauthorized assets for use in generated content.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The general instructions to search the web for logos, headshots, and templates describe external discovery behavior without clear disclosure, approval, or safety constraints. In this context, that can lead to privacy leakage, ingestion of malicious or copyrighted assets, and use of untrusted URLs in downstream generation workflows.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.