NYC
skills/aahl/skills/edge-tts-uvx/Gen Agent Trust Hub

edge-tts-uvx

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION] (HIGH): The skill uses uvx to execute shell commands where user-controlled input {msg} is interpolated directly into the command string. Evidence: uvx edge-tts --text "{msg}" .... Risk: An attacker can escape the --text argument and execute arbitrary commands using shell metacharacters.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill uses uvx to dynamically pull the edge-tts package from an external registry. Evidence: usage of uvx edge-tts in SKILL.md.
  • [PROMPT_INJECTION] (HIGH): The skill presents a significant indirect prompt injection surface (Category 8). 1. Ingestion points: {msg} placeholder in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Shell command execution via uvx. 4. Sanitization: Absent. The lack of delimiters or escaping allows malicious data to influence the agent's command execution environment.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 08:33 AM