ClawHub
Back to skill

Security audit

iA Presenter

Security checks across malware telemetry and agentic risk

Overview

The skill’s iA Presenter deck workflow is coherent, but its preview and theme helpers can affect local app state and capture the full screen without strong opt-in boundaries.

Install only if you are comfortable with a deck-authoring skill that can run local helper scripts, open iA Presenter, install or replace custom themes, and optionally capture screenshots. Ask the agent before any preview screenshot, preference change, or theme replacement, and prefer local media assets when you do not want deck rendering to contact third-party services.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding
The skill documents file read/write and shell-capable helper scripts, but does not declare permissions or clearly bound those capabilities. Undeclared execution and filesystem access increase the chance that a caller invokes actions with side effects the platform or user did not explicitly consent to, especially when the skill can open apps and manipulate local bundles.

Tp4

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding
The behavior described in the body exceeds the high-level declaration by including automatic app launching, theme enumeration from user directories, UI automation/activation, and screenshot capture of rendered previews. That mismatch is dangerous because users may authorize a content-authoring skill without realizing it can inspect local application data or capture screen contents, which can expose sensitive information beyond the deck itself.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill explicitly supports remote image URLs and YouTube/content fetching, but does not warn that opening or previewing a deck may trigger outbound network requests to third-party services. That can leak IP address, user agent, timing, and possibly presentation context, and may load untrusted remote media during rendering.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The automation notes instruct changing a user's application preference file and capturing the screen, which can alter user settings and access sensitive on-screen information. Although the text mentions restoring values and notes the Screen Recording requirement, it does not frame these actions with explicit consent, safety boundaries, or privacy warnings, so an agent could perform invasive actions unexpectedly.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.