ClawHub
Back to skill

Security audit

FMT高清动态图生成器 v6.0

Security checks across malware telemetry and agentic risk

Overview

The skill appears to support image generation, but it also tells the agent to run local Python and a hard-coded script from another installed skill, which needs review before use.

Install only if you trust the publisher and understand that this skill may execute local Python plus a generator script from another installed skill. Before large batch jobs, choose an output folder, confirm expected file counts, and review the referenced nano-banana-pro script separately.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Context-Inappropriate Capability

Medium
Confidence
87% confidence
Finding
The skill documentation instructs executing local Python code and invoking another skill's script from a filesystem path, which extends behavior beyond declarative image/video generation into arbitrary local code execution. Even though the embedded code is aimed at image post-processing, this pattern increases attack surface because skill content is untrusted and can normalize running shell/Python snippets with workspace and local filesystem access.

Context-Inappropriate Capability

Low
Confidence
80% confidence
Finding
The skill explicitly depends on another skill's local path (`~/.codex/skills/nano-banana-pro/scripts/generate_image.py`) without declaring that dependency as an allowed capability. This is dangerous because it couples execution to undeclared local resources and can be abused to reach unexpected code or encourage operators/agents to execute unreviewed filesystem-resident scripts.

Missing User Warnings

Medium
Confidence
76% confidence
Finding
The skill directs creation of multiple files and directories under `/workspace` in batch mode without explicit user-facing notice, consent, or limits. This can lead to unexpected disk usage, clutter, and side effects from broad requests like full-document generation, especially when combined with iterative regeneration and video outputs.

VirusTotal

56/56 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.