Security audit

Agentvibes Skill Free

Security checks across malware telemetry and agentic risk

Overview

This TTS voice skill is not clearly malicious, but it needs review because it asks for command execution while giving overbroad triggers and inconsistent offline/API-key guidance.

Install only if you are comfortable with a voice skill that can run commands, may automatically download TTS engines or voice files from HuggingFace, and can delete cached audio. Keep its use limited to TTS tasks, verify whether an API key is actually needed before entering any secret, and prefer a version that documents exact download sources and cache paths.

SkillSpector

By NVIDIA

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 (4)

Intent-Code Divergence

Medium

Confidence: 93% confidence
Finding: The skill documentation makes directly contradictory security-relevant claims: one section says the free version requires no API key, while the known limitations state that an API key is required and the skill cannot be used without one. This inconsistency can mislead users and downstream agents about authentication and data-flow expectations, increasing the chance of unsafe execution, failed runs, or accidental exposure of credentials if users try to supply secrets unnecessarily.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger condition says to use this skill for API integration, webhook configuration, and system connection tasks, which does not match a TTS voice synthesis skill's stated purpose. Overbroad or mismatched invocation guidance can cause agents to select this skill in inappropriate contexts and then use its exec capability on unrelated tasks, expanding attack surface and creating opportunities for unsafe command execution.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The documentation states that Piper automatically downloads voices from HuggingFace on first use, but the skill is presented as an offline/free local TTS tool without a prominent warning that it may initiate external network access. Hidden or underemphasized network behavior is dangerous because users and agents may assume local-only execution while the skill reaches external infrastructure and pulls unpinned content.

Missing User Warnings

Low

Confidence: 79% confidence
Finding: The cleanup command is documented as a cache-clearing action without clearly warning that cached audio files will be deleted. Even if limited to cache data, undocumented deletion behavior can cause unintended data loss, especially if users rely on generated audio artifacts for later reuse or if cache paths are broader than expected.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.