ClawHub
Back to skill

Security audit

Fantasy Transform Video Gen Seedance2.0

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed paid WeryAI video-generation skill with a broader underlying CLI, but no evidence of hidden data access, persistence, or deception.

Install only if you trust this WeryAI integration and are comfortable using a paid WERYAI_API_KEY. Before approving generation, check that the confirmation uses model SEEDANCE_2_0, review the expanded prompt, and avoid submitting confidential text or private image URLs.

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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The skill markets itself as tightly scoped to Seedance 2.0 fantasy transforms, but the documentation admits the underlying CLI supports broader model selection, more input modes, and unrelated operations. That gap weakens policy boundaries: an agent or user relying on the documented restrictions could be tricked into running broader, costlier, or less-reviewed operations than intended, including using different models or task modes with different safety properties.

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The file explicitly states the script requires a non-empty `model` but does not enforce the skill's claimed restriction to `SEEDANCE_2_0`. Relying on the caller to obey documentation creates a policy-bypass condition where downstream automation can select arbitrary supported models, defeating the skill's stated safety and behavioral constraints.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.