Outbound Call

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed outbound-calling helper that uses expected phone-call credentials and includes a consent requirement before placing calls.

Install only if you intend the agent to place outbound calls. Keep the ElevenLabs credentials scoped to this use, expect possible call charges, and require the agent to restate the destination number and get a clear yes immediately before every call.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The trigger phrases are broad, common-language requests like 'call someone' and 'phone someone,' which can match ordinary conversation and cause unintended activation. Because this skill places outbound phone calls through external services, accidental invocation can lead to privacy issues, unwanted contact, or financial cost, even though the skill text says to confirm consent first. The telephony context makes broad matching more dangerous than in a read-only or informational skill.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal