小米触屏闹钟视频制作

Security checks across malware telemetry and agentic risk

Overview

This skill is a visible Bilibili video download and processing guide; its main risks are ordinary media overwrite and package-install caution, not hidden or malicious behavior.

Install only if you are comfortable running local media-processing commands and downloading Bilibili content. Use a Python virtual environment where possible, avoid reusing output filenames, and keep originals until you confirm the cropped, trimmed, and compressed result is correct.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The skill repeatedly documents ffmpeg commands with the `-y` flag, which forces overwriting output files without prompting. Although the skill recommends BV-based filenames to reduce collisions, user-supplied paths, reused output names, or mistakes in intermediate/final file selection can still cause silent data loss, especially in a batch or scripted workflow.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal