Kokoro Agent Voices

• Outcome: Review
• Latest audit: May 29, 2026 at 9:29 PM UTC
• Version:
  1.0.1

Overview

The skill appears to provide text-to-speech playback, but its playback helper can turn a crafted output filename into a local shell command.

Review before installing or using. Avoid the --play option with any untrusted or unusual output path until playback is changed to use a safe subprocess argument list, such as passing afplay and the output filename as separate arguments.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (2)

os.system() or os exec-family call

High

• Category: Dangerous Code Execution

Content

out = speak(args.text, voice=voice, speed=speed, output=args.output)

    if args.play and out:
        os.system(f"afplay {out}")

if __name__ == "__main__":
    main()

• Confidence: 98% confidence
• Finding: os.system(f"afplay {out}")

Context-Inappropriate Capability

Medium

• Confidence: 84% confidence
• Finding: This finding points to the same risky behavior as the os.system call: local shell execution for playback. In context, playback is a plausible feature for a TTS tool, but implementing it through shell execution with unsanitized input makes it dangerous because it creates an avoidable command-injection path.

VirusTotal

64/64 vendors flagged this skill as clean.