ClawHub
Back to skill

Security audit

PDF PPTX Watermark

Security checks across malware telemetry and agentic risk

Overview

This is a local PDF watermarking skill whose behavior matches its stated purpose, with ordinary document-processing cautions around LibreOffice and broad trigger wording.

Install only if you are comfortable running local document-conversion tooling. Avoid feeding untrusted Office files unless LibreOffice is patched and isolated, confirm the skill is being used for watermarking rather than generic PDF conversion, and clean temporary conversion files if the source presentation is sensitive.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill documentation advertises shell execution and file access behavior but declares no corresponding permissions, creating a transparency and policy gap. In an agent environment, undeclared file-read and shell capabilities can cause users or orchestration layers to authorize a tool under false assumptions, increasing the risk of unintended local file access or command execution.

Tp4

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding
The documented purpose emphasizes real-time mobile tuning for watermarking, but the described behavior includes broader file-processing capabilities such as direct PDF handling and config-driven operation without an actual mobile interface. This mismatch is dangerous because capability expansion hidden behind a narrower description can mislead users, reviewers, or automated policy systems about what the skill really does and what data it can process.

Context-Inappropriate Capability

Medium
Confidence
79% confidence
Finding
The skill invokes LibreOffice to parse and convert user-supplied PPT/PPTX files. Even though the Python code avoids shell injection, handing untrusted documents to a large external office suite increases attack surface and could enable exploitation of converter/parser vulnerabilities or unintended file access in the converter process.

Vague Triggers

Medium
Confidence
85% confidence
Finding
The trigger phrases are broad, generic terms like 'convert to PDF' and 'add watermark' that can match many ordinary user requests, causing the skill to activate outside the user's specific intent. In an agent setting, over-broad activation can route unrelated documents into a file-processing workflow, increasing the chance of unintended file handling or tool invocation.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.