ClawHub
Back to skill

Security audit

Multimodal Asset Tagger

Security checks for vulnerabilities and agentic risk

Overview

This skill is low-risk but appears to overpromise multimedia metadata features that its bundled script does not actually implement.

Install only if you want basic guidance for AI-friendly media metadata and simple image alt-text/filename generation. Expect incomplete functionality for video, audio, captions, schema markup, keywords, and the documented --output option until the author aligns the script with the documentation.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (4)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The declared description promises a broader multimedia optimization capability: alt text, file names, captions, and Schema markup for images, videos, and audio assets. The actual code implements only a narrow subset: basic alt text and filename generation for an image-like asset. The --type parameter suggests multimedia support, but it is not used in behavior, making the declared image/video/audio coverage materially overstated. This is a description-behavior mismatch because the primary scope and outputs are significantly narrower than claimed.

Vague Triggers

Medium
Confidence
94% confidence
Finding
The skill description uses very broad activation phrases such as 'use whenever the user mentions optimizing images for AI' and similar variants, which can cause the agent to invoke this skill in many loosely related contexts. Over-broad triggering increases the chance of inappropriate tool selection, irrelevant instruction injection into benign conversations, and unintended processing of user content under the wrong workflow.

Description-Behavior Mismatch

Medium
Confidence
98% confidence
Finding
The manifest describes a multimodal tagging skill that generates alt text, file names, captions, and Schema markup for image, video, and audio assets. In this file, the implementation only formats a text string into image-style alt text and a .jpg filename, with no handling for video/audio outputs, captions, or Schema markup.

Intent-Code Divergence

Medium
Confidence
96% confidence
Finding
The interface and module wording suggest multimedia-aware behavior, yet main() unconditionally calls optimize_image regardless of the --type argument. This creates an intent-versus-implementation contradiction because users are led to believe type-specific processing exists when it does not.

Static analysis

No suspicious patterns detected.