Create Content

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only writing skill that fits its stated purpose, though users should limit any access to personal notes or past sessions.

Install is reasonable for a writing assistant. Before using the idea-discovery flow, tell the agent exactly which notes, journals, folders, or time range it may review, exclude confidential or highly personal material, and review drafts carefully before posting.

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 example invocation includes the phrase "help me figure out what to post," which is natural conversational language a user could say outside an intentional tool invocation. In agent systems that route on broad phrase matching, this can cause accidental skill activation and unintended access to the skill's search-and-drafting behaviors.

Vague Triggers

Medium

Confidence: 97% confidence
Finding: The shortcut triggers "explore" and "help me think" are extremely generic phrases that commonly appear in normal conversation. If the host agent uses these as activation cues, the skill may hijack unrelated interactions, causing prompt-routing errors and unintended behavior in contexts far outside content creation.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal