Back to skill

Security audit

PDF Watermark

Security checks across malware telemetry and agentic risk

Overview

This skill does what it claims: locally converts PPT/PDF files and writes a watermarked PDF, with a few normal privacy cautions around local processing and the optional web tunnel.

Install only if you are comfortable running a local Python document-processing script and LibreOffice on the files you choose. Avoid using the optional public tunnel with sensitive material, and check or clean temporary directories after converting confidential PPT/PPTX files.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill instructs users to run local Python scripts, invoke LibreOffice in headless mode, and read/write local files, but the metadata declares no permissions or equivalent trust signals. This creates a transparency and review gap: users or platforms may underestimate that the skill executes shell commands and processes local documents, increasing the risk of unsafe use or insufficient sandboxing.

Vague Triggers

Medium
Confidence
86% confidence
Finding
The trigger phrases are broad, generic terms like 'convert to PDF' and 'add watermark' that can match many normal user requests and cause the skill to activate outside a narrowly intended scope. In an agent environment, over-broad activation can lead to unintended file handling or tool invocation, increasing the chance of acting on sensitive documents without explicit user intent for this specific skill.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill suggests exposing a locally served UI through a public tunnel but does not warn that anyone with the generated URL may access the interface and any content it serves. Even if the UI is only for parameter tuning, publishing a local HTTP server can leak local files from the served directory or enable unintended access, especially since `python3 -m http.server` serves the current directory contents.

Missing User Warnings

Low
Confidence
92% confidence
Finding
The code creates a temporary directory with tempfile.mkdtemp() for converted PDFs and never removes it. This can leave residual converted documents on disk, which may expose sensitive file contents to other local users or cause data retention and disk-usage issues over time.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.