Lux Tts

Security checks across malware telemetry and agentic risk

Overview

This looks like a local TTS skill, but it recommends running missing administrator-level installer scripts and the included implementation appears to be a simulated TTS rather than the advertised full voice-cloning model.

Review carefully before installing. Do not run the referenced install.bat, deploy.ps1, or download scripts as administrator unless you have obtained and inspected them from a trusted source. Prefer a manual virtual environment with pinned dependencies, verify the Hugging Face model source, and be aware that the packaged code appears to be a mock TTS implementation rather than the advertised full voice-cloning system.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI04: Agentic Supply Chain Vulnerabilities
Medium
What this means

Following the recommended setup could run installer code that was not included in the reviewed skill package.

Why it was flagged

The guide recommends an administrator one-click installer, but the provided file manifest does not include install.bat or the referenced deploy/download scripts, leaving users dependent on unreviewed external or local code.

Skill content
选项 1:一键安装(推荐) ... # 以管理员身份运行 ... E:\桌面\openclaw-main\workspace\lux-tts\install.bat
Recommendation

Do not run missing installer scripts blindly; obtain them from a trusted source, inspect them first, and prefer a pinned manual virtual-environment install.

#
ASI03: Identity and Privilege Abuse
Medium
What this means

An administrator installer can modify the system broadly if the referenced script is unsafe or replaced.

Why it was flagged

Administrator execution is requested for installation, but the artifacts do not justify why a local TTS integration needs elevated privileges, and the referenced script is not provided for review.

Skill content
# 以管理员身份运行 ... E:\桌面\openclaw-main\workspace\lux-tts\install.bat
Recommendation

Avoid administrator execution unless absolutely necessary and only after reviewing the exact installer contents.

#
ASI05: Unexpected Code Execution
Medium
What this means

A user could be induced to run an unreviewed PowerShell script with reduced execution-policy protections.

Why it was flagged

The guide suggests bypassing PowerShell execution policy for a deployment script that is not included in the reviewed artifacts; this is user-directed setup, but it weakens a local script-execution safeguard.

Skill content
powershell -ExecutionPolicy Bypass -File deploy.ps1
Recommendation

Use normal execution policy where possible, inspect the script, verify its source, and avoid bypass flags for unknown setup files.

#
ASI09: Human-Agent Trust Exploitation
Low
What this means

Users may trust the skill to provide real high-quality voice cloning when the packaged implementation may only provide a mock/test generator.

Why it was flagged

The code discloses that the included ready implementation is a simulated version generating mock audio, while the skill description advertises a high-quality LuxTTS model with voice cloning.

Skill content
完全可用的 LuxTTS 模拟版本 ... 未来可以无缝替换为真实模型 ... # 生成模拟音频
Recommendation

Clearly label the skill as a simulator or include the real model integration, and avoid relying on it for production TTS until verified.

#
ASI02: Tool Misuse and Exploitation
Low
What this means

Incorrect or unsafe paths could copy or overwrite local files in the voice directory.

Why it was flagged

The voice-management helper copies user-selected audio files into the persistent voices directory; this is purpose-aligned for voice cloning, but path/name choices should remain user-controlled and reviewed.

Skill content
target_path = os.path.join(voices_dir, name) ... shutil.copy2(audio_path, target_path)
Recommendation

Only add trusted voice files and use simple filenames rather than absolute or traversal-style paths.