Skills
skills/inference-shell/skills/ai-voice-cloning/Gen Agent Trust Hub

ai-voice-cloning

Pass

Audited by Gen Agent Trust Hub on Apr 22, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external URLs and has the capability to execute shell commands.
  • Ingestion points: Parameters such as video_url, image_url, and audio_url used in the media-merger and animated avatar workflows.
  • Boundary markers: Not present; the skill lacks instructions for the agent to ignore or isolate content within retrieved external files.
  • Capability inventory: The skill is authorized to use the Bash(infsh *) tool, allowing it to perform various media processing and generation tasks.
  • Sanitization: No sanitization or validation of content from external URLs is specified.
  • [COMMAND_EXECUTION]: The skill executes infsh CLI commands to interact with AI voice models and merge media files. These operations are consistent with the skill's primary purpose and are correctly restricted within the allowed-tools configuration.
  • [SAFE]: References to installation scripts and related skills point to official resources within the inference-sh GitHub organization and domain. These are treated as legitimate vendor-owned resources.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 22, 2026, 04:25 PM
Security Audit — agent-trust-hub — ai-voice-cloning