Draft Docs

Security checks across malware telemetry and agentic risk

Overview

This skill coherently helps draft and optionally publish documentation, with the publishing behavior disclosed and user-directed.

Review drafts before using --publish, and confirm the destination and navigation changes because publishing can make documentation visible in your site. No evidence suggests hidden data access, exfiltration, or destructive behavior outside the docs workflow.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
The frontmatter description says the skill generates first-draft documentation, but the body also supports a publish mode that moves files into live docs locations and updates navigation. This mismatch can mislead users or higher-level tooling into invoking a skill with broader side effects than advertised, increasing the chance of unintended publication changes.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The publish workflow performs state-changing operations that can make documentation live by moving files and editing navigation, yet it lacks a prominent user-facing warning that these actions affect published content. In this context, the skill is more dangerous because documentation trees and navigation files are production-facing artifacts, so an accidental invocation or ambiguous destination choice can cause unauthorized or unintended site changes.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal