create-content

Security checks across malware telemetry and agentic risk

Overview

This is a mostly coherent content-writing skill, but it asks the agent to search personal notes, journals, and prior sessions for public post ideas without clear limits.

Install only if you are comfortable with the agent using personal notes, journals, or prior sessions to brainstorm content. Before using it, specify exactly which documents or sources may be searched, exclude sensitive journals or private conversations, and review drafts carefully before posting publicly.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The invocation example includes a broad natural-language phrase, specifically "help me figure out what to post," which overlaps with ordinary conversation and could cause unintended skill activation. In an agent environment, ambiguous triggers increase the chance that normal user text is interpreted as a command, leading to incorrect routing, unexpected tool behavior, or prompt-injection exposure through accidental activation.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The quick commands define short, everyday phrases like "explore" and "help me think" as activation triggers without strong boundaries. These are common utterances in normal conversation, so they create a realistic risk of accidental invocation or command confusion, especially in multi-skill or agentic systems where overlapping phrases may route control to the wrong skill.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal