ClawHub
Back to skill

Security audit

Diagram生成器

Security checks across malware telemetry and agentic risk

Overview

The skill mostly matches a diagram-generation purpose, but it needs review because it asks agents to run an external connector and write or overwrite files while some instructions under-disclose those powers.

Review before installing. Use only if you trust the connector package, pin or verify the command source where possible, keep outputs in a dedicated workspace directory, and do not provide callback URLs or sensitive diagram content unless you understand where it will be sent.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The skill presents itself at the end as a pure Markdown skill while earlier declaring `exec`, `write`, and connector-driven command execution via `npx`. This mismatch can mislead users and host agents into applying weaker trust assumptions, causing command execution or file-modifying behavior to occur when the operator expects documentation-only behavior.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill repeatedly describes generating files, creating directories, and overwriting outputs, including edit-mode overwrite behavior, but does not provide a clear up-front warning that local files may be created or replaced. In an agent environment with write access, this can lead to unintended modification or destruction of user files, especially when filenames or output paths are supplied dynamically.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The request format includes a `callback_url` for asynchronous completion, but the skill does not prominently warn that user data, diagram specs, filenames, or metadata may be transmitted to an external endpoint. This creates a real exfiltration and SSRF-style risk in agent environments if an attacker can influence the callback URL or payload contents.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.