ClawHub
Back to skill

Security audit

ZenStudio CLI

Security checks across malware telemetry and agentic risk

Overview

This ZenStudio skill is mostly coherent, but it gives agents broad account-changing CLI authority, including an arbitrary MCP tool passthrough, without clear guardrails.

Install only if you trust the ZenStudio npm CLI and intend to let an agent manage ZenStudio content with your API key. Prefer normal named commands, confirm before uploads, downloads, canvas delete/overwrite actions, or `tools call`, and protect the saved `~/.zencli` credentials.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
The skill is presented as a ZenStudio content-creation CLI, but it also documents a generic `zencli tools call <tool_name>` capability that can invoke arbitrary MCP tools beyond the narrowly described scope. This creates a scope-expansion risk: an agent may treat the skill as safe for image/video/project operations while actually being able to trigger broader backend actions, increasing the chance of unintended privileged operations or misuse of sensitive integrations.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The invocation guidance is broad enough to match generic requests like uploading files, downloading assets, or managing materials without clearly limiting those actions to the ZenStudio platform context. This can cause an agent/router to invoke the skill for ordinary file-transfer tasks the user did not intend to perform through ZenStudio, potentially causing data disclosure, accidental uploads, or actions against the wrong external service.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.