Security audit

Agentvibes Voice Skill Free

Security checks for vulnerabilities and agentic risk

Overview

This TTS skill appears non-destructive, but its documentation asks for broader execution, file, API, and credential-related authority than a Piper-only voice skill clearly needs.

Review this skill carefully before installing. It is probably a rough or template-heavy TTS skill rather than clearly malicious, but users should restrict it to Piper text-to-speech tasks, avoid providing API keys unless the publisher explains why they are needed, and be aware that first-time voice use may download files from HuggingFace.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Intent-Code Divergence

Medium

Confidence: 92% confidence
Finding: The skill presents itself as a free, offline Piper-only TTS client, but later introduces API key setup and remote-service style failure modes that materially change its trust and data-flow model. This can mislead an agent or user into invoking networked or credentialed behavior they did not consent to, increasing the risk of secret exposure or unintended outbound communication.

Intent-Code Divergence

Low

Confidence: 81% confidence
Finding: The documentation states the free version only supports Piper, yet includes provider-management commands that imply broader provider handling. That inconsistency can cause an agent to probe unsupported providers or execute broader environment checks than necessary, expanding behavior beyond the advertised scope.

Intent-Code Divergence

Medium

Confidence: 95% confidence
Finding: A skill advertised as TTS-only later claims generic file handling, API integration, and command execution capabilities, which is a significant scope expansion. In an agent environment with read/write/exec tools, this kind of overbroad capability description can trigger unintended privileged actions unrelated to TTS and mask a more dangerous operational surface.

Vague Triggers

High

Confidence: 96% confidence
Finding: The activation guidance uses very broad 'Use when' language spanning unrelated domains like API integration, webhooks, system connection, video/audio processing, and media conversion. Overbroad invocation criteria can cause an agent to select this skill in inappropriate contexts, effectively creating privilege and scope creep that may lead to unsafe exec, file, or network actions.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The parameter description for input is so vague that it allows nearly any 'data or instruction' to appear in scope. In an agent setting, ambiguous inputs make it easier for unrelated or unsafe instructions to be funneled into a skill that has exec/write capabilities, increasing the chance of misuse or prompt-to-action confusion.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill is described as free/offline, but the markdown does not clearly warn at the point of use that first-time voice selection downloads files from HuggingFace. This hidden network dependency undermines user expectations about offline behavior and can expose systems to unreviewed external downloads or policy violations.

Static analysis

No suspicious patterns detected.