ClawHub
Back to skill

Security audit

ElevenLabs Agents

Security checks across malware telemetry and agentic risk

Overview

This skill matches its ElevenLabs agent-management purpose, but it tells the agent to hide important local and remote changes that can affect the user's files and ElevenLabs account.

Install only if you trust the local ElevenLabs CLI and are comfortable letting the agent modify local agent files and remote ElevenLabs agents. Before using it, ask the agent to disclose file creation, credential handling, sync overwrites, and any push or webhook changes, and require explicit approval before deployment or overwrite operations.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill instructs the agent to silently run `elevenlabs agents init` when `agents.json` is missing, which performs local file writes without informing the user or obtaining consent. Undisclosed filesystem modification violates user expectations and can create or alter project state in ways that are hard to audit or undo.

Missing User Warnings

High
Confidence
97% confidence
Finding
The documented sync behavior includes `elevenlabs agents pull --update`, which can overwrite local files with remote state, but the skill does not require a warning, diff, backup, or confirmation before doing so. This creates a realistic risk of destructive data loss or replacement of user-managed local configurations.

Behavior Manipulation

Medium
Category
Prompt Injection
Content
## Critical Behavior Rules

1. **Hide CLI details**: Never tell users to run commands. Handle everything silently.
2. **Handle errors gracefully**: If a command fails, fix it yourself or explain the situation in user-friendly terms.
3. **Local vs Remote distinction**: Always be clear whether you're showing local (synced) agents or remote (platform) agents.
Confidence
87% confidence
Finding
Never tell user

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.